ACCOUNT ACCESS
Manuscripts can be permanently deleted while retaining generated reports.
VERIFY YOUR EMAIL
We sent a 6-digit code to . Enter it below to activate your account. It may take a few minutes — please check your spam folder.
BETA ACCESS CONFIRMED
SHARE REFEREE
Know a researcher who could use private pre-submission peer review? Share with your lab, coauthors, or grad students — they get a free review.
BUILT FOR LIFE-SCIENCE RESEARCHERS
A private pre-submission review built for life-science authors: surface the objections, journal-fit gaps, and revision priorities that cost teams time after submission.
Every account receives a complimentary Starter tier for one month — 3 reviews, 3 revision checks, and 3 manuscript edits. No credit card required. Feedback is always welcome and helps us improve.
FROM DRAFT TO SUBMISSION
THE NEW REVIEW ROOM
RefereeBio gives life-science teams a private way to pressure-test a manuscript before submission: likely critiques, statistical objections, journal-fit concerns, and the strongest next move.
A 2026 Nature Portfolio editorial reports that 53% of peer reviewers in a global survey had used AI tools in their work. Read the source.
WHAT YOU GAIN
Get ranked revision priorities alongside a credible journal path for the draft you have now.
Referee compares your manuscript across a catalog of 150+ life-science journals, then anchors the review to each venue’s expectations for novelty, mechanism, rigor, scope, and audience. You see a reach, best-fit, and fallback path—not just a prestige list.
Identify statistical and methodological weaknesses before reviewers do.
It translates methods, figures, and reporting choices into reviewer-facing risk: replicate definitions, correction choices, effect sizes, pseudoreplication traps, overclaiming, and missing limitations.
Upload reviewer comments privately and assess whether the revised manuscript has adequately addressed each concern.
Referee tracks major and minor reviewer concerns against your revised manuscript, identifies what is adequately addressed, and flags what still needs work before you resubmit.
LIFE SCIENCE, NOT GENERIC AI
Referee is designed around the decisions life-science teams actually make: whether the evidence supports the claim, what a target venue will expect, and which revision is worth doing before the manuscript enters a slow and expensive review cycle.
Read the FAQEXPLORE THE OUTPUT
Choose a card to preview the part of Referee that matters most to you. The gallery advances automatically, but you remain in control.
SIMULATED EDITORIAL RISK ESTIMATE
Critical gap: replicate structure and effect-size reporting may trigger reviewer concern.
Choose a journal sequence instead of guessing one venue at a time.
Our data prove that TIA-1 condensation drives stress-granule nucleation. Our data indicate that TIA-1 condensation is associated with stress-granule nucleation.
Thus, modulating TIA-1 phase behavior represents a promising strategy. Future work is needed to assess whether modulating TIA-1 condensation directly perturbs stress granule assembly.
REVISION MOMENTUM
75%+24% since first review51% 63% 67% 67% 75%DESIGN PRINCIPLES
The most useful review identifies the two or three objections most likely to change the outcome—not every possible weakness.
It is more useful to surface a statistical or mechanistic gap before submission than after peer review begins.
Referee does not promise publication outcomes. It gives authors a clearer map of risk, leverage, and next steps.
Questions about how Referee works or its limits? Read the FAQ.
END-TO-END MANUSCRIPT INTELLIGENCE
One upload activates the full suite. No juggling tools. No manual handoffs between review, editing, and resubmission.
Three expert reviewer personas — with distinct methodological, statistical, and conceptual lenses — each write a differentiated report anchored to your target journal's standards.
Identifies statistical reporting gaps — replicate definitions, effect sizes, correction choices, pseudoreplication — before a reviewer uses them to question your rigor.
Directional desk-triage signals plus simulated outcome probabilities — desk reject, review, major revision, minor revision — contextualized to your chosen venue.
Searches PubMed and OpenAlex for published work that substantially overlaps your claims. High-overlap results surface as novelty warnings before reviewers raise them.
Sentence-level screening for formulaic, repetitive, or unusually uniform language associated with AI-generated prose. Each flag explains the pattern and suggests a clearer, more natural alternative. Results are probabilistic and do not establish AI authorship.
Matches your manuscript against 150+ life-science journals and returns a reach, best-fit, and fallback cascade ranked by scope, novelty bar, rigor standards, and fit score.
Produces a polished, journal-specific cover letter — with the correct scope framing, complementary-paper callouts, and editorial conventions for your chosen venue.
Track-changes suggestions with inline flags for overclaiming, hedging failures, scope creep, and unclear language — downloadable as an annotated DOCX.
Auto-drafts a structured response to reviewer comments — one reply per concern, in the standard rebuttal format, ready to edit before you resubmit.
Upload your revised manuscript and reviewer comments together. Referee tracks which concerns are addressed, which need more work, and what the delta is from the original.
A journal-specific readiness score that aggregates statistical completeness, novelty, scope fit, and reviewer-risk to tell you whether the manuscript is ready to send.
My Papers hub keeps every review, revision cycle, score trajectory, and exported file in one place — with momentum tracking across your full submission history.
Scope: Referee currently supports primary research articles in the life sciences — biology, biochemistry, biophysics, computational biology, chemical biology, pharmacology, structural biology, genetics, and related disciplines. Support for literature reviews, meta-analyses, and other article types is planned for a future update.
BEFORE THE DECISION LETTER
A simulation for manuscript preparation—not an editorial decision or a publication forecast.
FROM DRAFT TO SUBMISSION
THE NEW REVIEW ROOM
As AI-assisted scholarly workflows become more common, authors deserve a disciplined way to pressure-test a manuscript before submission. RefereeBio gives authors a private simulation of critiques, statistical objections, and journal-fit concerns that may surface during review.
A 2026 Nature Portfolio editorial reports that 53% of peer reviewers in a global survey had used AI tools in their work. Read the source.
WHO IT IS FOR
Prepare a submission with a clearer view of desk-rejection risk, likely reviewer objections, and whether the target journal is worth the attempt.
Avoid preventable reviewer criticism by seeing the fixes that matter first, especially around claims, figures, methods, and statistical reporting.
Decide between reach, best-fit, and cascade journals with a structured view of scope, novelty, audience, and revision burden.
BEFORE YOU SUBMIT
Sample report signal"Likely reviewer concern: statistical analysis plan is under-specified for subgroup comparisons."
WHY REFEREE IS DIFFERENT
Most AI manuscript tools help polish prose, summarize literature, or suggest generic improvements. RefereeBio is built around the decision environment: journal fit, reviewer psychology, statistical reporting, desk triage, and the specific objections that can derail a paper after months of waiting.
Feedback is interpreted against venue expectations rather than a universal checklist.
Multiple reviewer archetypes expose conflicting pressures that a single generic AI response misses.
The report checks methods, legends, and reporting signals that reviewers often use to infer fragility.
Instead of giving a long undifferentiated list, Referee identifies what changes the outcome first.
Reach, best-fit, fallback, and related journals are framed as a sequence, not a vanity list.
Turn critique into a PDF, DOCX, or TXT package for lab meetings, coauthors, and revision planning.
HOW IT WORKS
Paste text or upload a PDF, DOCX, or TXT. Select your target journal and research field. That's it — no formatting required.
Three simulated reviewer perspectives score your manuscript against the selected journal's context and write differentiated reports. You also get a statistical methods audit, section-by-section risk flags, and directional desk-triage signals.
Use the ranked fix list to prioritize. The cover letter draft and submission cascade tell you where to send it. If you get reviews back, use the Revision check tool for a point-by-point response framework and revision assessment.
SEE IT IN ACTION
This is a fabricated walkthrough on an invented manuscript, shown to illustrate the output. Every score, signal, and comment below is generated the moment you upload — calibrated to the journal you select.
Sample only. The manuscript, journals, scores, and reviewer language shown here are fictional and for illustration. Journal names are used descriptively for submission-planning context; RefereeBio is independent and not affiliated with, endorsed by, or operated by any journal, publisher, editor, or society.
Referee turns a manuscript into a practical risk readout: likely revision burden, best-fit journal, and the outcome signals that help you decide whether to revise, redirect, or proceed.
This is a simulation for pre-submission manuscript preparation. It is not an official editorial decision, does not impersonate a journal or editor, and should not be treated as a publication forecast.
Scores separate novelty, mechanism, statistics, figures, and scope so the team can focus on the parts reviewers are most likely to punish.
The audit translates methods and figure legends into reviewer-facing risk: replicate definitions, correction choices, effect sizes, and pseudoreplication traps.
Paper type: Cell-biology imaging study with live-cell microscopy, FRAP recovery kinetics, and automated quantification of stress-granule number per cell.
Specify whether granule counts and FRAP measurements are biological or technical replicates, adopt a mixed-effects model that nests cells within independent experiments, and report 95% CIs on FRAP half-times alongside point estimates.
The cascade frames submission as a route: reach, best-fit, fallback, and what to do next if the first decision does not go your way.
Journal selection is based on primary topic, rigor, novelty, and audience — not tools or methods used. Consider Best Fit first; if rejected, move to the next tier rather than a lateral same-prestige journal with different scope.
Reach
High-impact cell biology with strong demand for mechanism and physiological relevance.
Best Fit
Best Fit reflects both strong topical alignment and the optimal selectivity tier for this manuscript's current form — not simply the highest fit score. A higher-fit reach journal may have a lower score here because its selectivity bar exceeds the manuscript's current strength.
Rigorous cell biology with emphasis on quantitative imaging, mechanism, and data quality.
Fallback
Transparent, constructive review with appetite for methodologically careful cell-biology studies.
For a higher-tier target, Referee identifies the few changes most likely to move the editorial needle instead of burying you in generic advice.
To reach Nature Cell Biology, the manuscript needs a causal mechanistic link between TIA-1 condensation and a functional cellular outcome, plus validation in a second physiological context.
These are the 3 specific changes most likely to close the gap between your submission journal and the reach journal. Ranked by likely impact on desk-triage risk.
Move beyond correlation between TIA-1 phase separation and granule number toward a causal test — for example, a separation-of-function mutant that abolishes condensation without disrupting RNA binding. The abstract should open with a mechanistic principle, not a phenotype.
Why Nature Cell Biology requires this: Nature Cell Biology weights conceptual and mechanistic advance more heavily than Journal of Cell Biology — closing this gap is the single highest-leverage change for the next selectivity tier.
Extend beyond the U2OS cell model to at least one primary or in vivo context where oxidative stress is physiologically relevant — for instance, primary neurons or a tissue oxidative-stress paradigm.
Why Nature Cell Biology requires this: Nature Cell Biology expects multi-system evidence for a central mechanistic claim rather than a single immortalized cell line.
Replace pooled-granule FRAP with per-cell nested statistics and report effect sizes with confidence intervals; add a dose-response across oxidative load rather than a single concentration.
Why Nature Cell Biology requires this: Reviewers at this tier read quantitative rigor as a proxy for whether the central mechanistic claim will hold up.
The final layer converts reviewer risk into concrete language changes, margin comments, and exportable edits your coauthors can act on.
Oxidative stress triggers the rapid assembly of cytoplasmic stress granules, membraneless organelles that sequester translationally stalled mRNAs alongside a network of RNA-binding proteins.1–4 TIA-1, a prion-like-domain protein, has been repeatedly identified as a core scaffold of these granules across human cell lines and primary cultures.5,6
Our data prove that TIA-1 condensation drives stress-granule nucleation Our data indicate that TIA-1 condensation is associated with stress-granule nucleation, with granule number scaling with the fraction of TIA-1 partitioned into the condensed phase.7 While it remains unclear whether condensation is a cause or a consequence of translational arrest, granule burden offers a tractable readout Whether condensation is a cause or a consequence of translational arrest remains unresolved. Regardless, granule burden offers a tractable, indirect readout of the cellular response to oxidative load. Thus, modulating TIA-1 phase behavior represents a promising strategy for tuning the stress response.F04 However, current approaches to perturbing condensation in living cells remain limited, and the field still lacks a separation-of-function tool that decouples assembly from RNA binding.8,9
PRIVATE
Use Referee before circulating a draft, before choosing a journal, or before a high-stakes resubmission. The goal is to expose the objections while they are still inexpensive to fix.
SPECIFIC
The same paper can look ambitious at one venue and underpowered at another. Referee frames critique against the target journal's likely tolerance for novelty, mechanism, statistics, and scope.
The product uses a journal catalog with venue-specific priors, reviewer weighting, and submission-context expectations.
Referee samples a weighted archetype pool, from constructive mentors to statistical enforcers and high-bar gatekeepers.
Reviewer mood, panel composition, and a calibrated Reviewer 2 effect model the kind of variability authors often experience in peer review.
WHAT THE MODEL CONSIDERS
RefereeBio scores the complete manuscript using journal-independent criteria such as novelty, conceptual depth, mechanism, technical strength, statistical confidence, figure strength, claim–evidence alignment, study scale, and methodological robustness. Journal culture is applied separately to reviewer commentary, editorial risk, revision requests, and journal fit.
The output is framed as simulated editorial risk: desk-triage vulnerability, post-review revision risk, and likely revision burden. It also ranks what to fix first so authors can improve the manuscript before submission. These are directional strategy signals, not actual journal outcome probabilities.
EXAMPLE PRIMARY REPORT
See the kind of primary manuscript report Referee generates: score criteria, reviewer-style objections, journal fit, ranked fixes, and submission strategy in one scrollable example.
REVISION MOMENTUM
Run the draft, revise the high-yield weaknesses, then watch your manuscript score improve as each version gets closer to submission-ready.
Start with my draftBEFORE THE DECISION LETTER
Analyzing your manuscript...
ANALYSIS IN PROGRESS
PRIVATE MANUSCRIPT INTELLIGENCE
Three simulated reviewers. Journal-calibrated risk signals. A ranked fix list. Ready before you submit.
FAQ
RefereeBio is a pre-submission preparation tool. These answers summarize the most important privacy, affiliation, and responsible-use limits.
WHAT REFEREE ANALYZES & PREPARES
Each report assesses the manuscript against the expectations of its journal context, then turns the most important risks into clear next steps.
Your manuscript is used to generate your private report, saved history, exports, and requested editing or revision check features. RefereeBio does not publish your uploads, submit them to journals, or sell manuscript content.
RefereeBio does not intentionally use private manuscripts to train public AI models. Manuscript text may be sent to configured AI providers to generate requested output, subject to provider terms and data-processing settings.
Referee does not make publication decisions, contact editors, submit manuscripts, certify statistics, replace collaborators or compliance review, or promise that a revision will change any journal outcome.
No. Journal and publisher names are used descriptively for submission-planning context only. RefereeBio is independent and is not affiliated with, endorsed by, sponsored by, certified by, or operated by any journal, publisher, editor, society, or indexing service.
No, not by default. Do not upload protected health information, patient-identifying data, HIPAA-covered data, FERPA-covered records, export-controlled data, classified data, or other specially regulated content unless RefereeBio signs a separate written agreement covering that data.
Yes. Uploaded manuscript files can be deleted while generated reports remain in your account history. Some operational records may remain temporarily in backups, logs, payment records, provider systems, or security records where required.
UNDERSTANDING YOUR RESULTS
Referee reports several different signals. They answer different questions and should not be read as interchangeable grades or publication probabilities. Read the methodology for a full explanation of each dimension.
This is a journal-independent assessment of the manuscript in its current form. Each dimension is built from specific, checkable evidence found directly in the manuscript's own text — not a single holistic impression — so the score reflects concrete, identifiable features of the manuscript rather than an overall vibe.
The assessment first identifies what kind of study this is — a mechanistic experiment, a genetic or chemical screen, a computational method, a clinical trial, a structural study, and so on — and evaluates the relevant dimensions against evidence standards appropriate to that study type. A screening paper isn't penalized for lacking a rescue experiment it was never designed to run; a computational method isn't judged by wet-lab standards. Many manuscripts blend more than one study type, and the assessment accounts for that blend rather than forcing a single category.
The ten dimensions carry different weight, with statistical confidence, claim–evidence alignment, and methodological robustness weighted most heavily; figure strength contributes when figures are available. Scores are calibrated against real manuscripts so the numbers mean something comparable across very different papers: a genuinely average, competent manuscript should land in the mid-70s, work strong enough for a well-regarded but non-flagship journal typically lands in the low-to-mid 80s, and only work that is exceptional across nearly every evaluated dimension reaches the 90s. It is not an acceptance probability, journal percentile, or editorial decision.
The quality assessment reads the complete extracted manuscript, including later results, methods, legends, supplements, limitations, and statistical details. An identical normalized manuscript reuses the same versioned assessment regardless of the journal selected; journal choice changes the reviewer commentary and journal-fit analysis, not this score.
Scope and audience alignment are reported separately in the journal-fit analysis and do not change manuscript quality.
Journal fit is journal-dependent and separate from manuscript quality. It combines relevance to the journal's scope and readership with whether the manuscript's novelty, mechanistic depth, statistical rigor, evidence base, and study scale meet that venue's expectations. It is therefore neither a pure topic-relevance score nor the manuscript's overall score, and it is not an acceptance probability.
Reach is a credible stretch target. Best Fit balances topical alignment with the most appropriate selectivity level for the manuscript's current strength. Fallback offers a more forgiving route. Because these labels represent submission strategy, Best Fit is not always the journal with the highest raw fit percentage.
This score measures how fully the revised manuscript and response address the original reviewer comments: addressed concerns receive full credit, partially addressed concerns receive partial credit, and unresolved concerns receive none. Major concerns drive the displayed assessment when available.
It is primarily dependent on the original reviewer comments and the evidence in the revision—not the journal itself—although those comments may naturally reflect that journal's standards. It does not replace the revised manuscript's fresh quality score: a revision can address every comment yet still have broader weaknesses, or materially strengthen the paper without answering every comment.
How to use the scores: compare dimensions to identify the manuscript's limiting factors, use journal fit to plan where to submit, and use revision improvement to check response coverage. Treat changes across versions and the written reviewer reasoning as more informative than any single number.
EXAMPLE PRIMARY REPORT
This fictional RefereeBio report demonstrates the structure of a generated pre-submission review: editorial risk, methods checks, journal fit, reviewer reports, ranked fixes, and writing diagnostics. It is not a real journal decision, endorsement, or publication forecast.
Fictional sample report. The manuscript scenario, journal context, scores, reviewer comments, prior-art examples, AI-detection notes, and cover letter are invented for product demonstration.
This is a simulation for pre-submission manuscript preparation. It is not an official editorial decision, does not impersonate a journal or editor, and should not be treated as a publication forecast.
Directional outcome simulation based on high-tier journal stringency, published selectivity priors where available, and this manuscript's quality scores.
Fictional sample report. Directional simulation only. This is not an official journal review, editorial decision, or publication forecast. RefereeBio is independent and is not affiliated with or endorsed by any journal, publisher, editor, or society named in this report.
Use this as a pre-submission risk map. The strongest signal is not the exact percentage; it is the pattern of objections, ranked fixes, methods gaps, and journal-fit pressure points that repeat across the report.
Referee cannot know the actual editor, reviewer panel, confidential journal priorities, competing papers under review, informal reviewer biases, or final decision. Treat the output as preparation, not prophecy.
Paper type: In vitro cell-based CRISPR screening study with mechanistic validation, protein biochemistry, and patient-derived cell modeling.
Holm and Holm-Sidak corrections applied to planned comparisons; Benjamini-Hochberg FDR used for screen-level enrichment; Dunnett correction for post hoc biochemical assays.
TEM scoring used randomized image identifiers by a blinded analyst; SPR injections and immunofluorescence imaging were performed with blinded or randomized order.
Comparisons specify named tests including repeated-measures ANOVA, Kruskal-Wallis, robust Z-scores, nonlinear fitting, bootstrap CI, SPR fitting, and mixed-effects models.
The manuscript distinguishes biological replicates from technical units and states that single-cell analyses treat biological replicate as the experimental replicate.
Major comparisons report 95% confidence intervals on effect sizes for imaging, rescue, and binding experiments.
Sample sizes are selected prospectively on practical and precision grounds and justified by effect-size precision and sampling depth.
Critical gaps: Add explicit confirmation that image-analysis thresholds were locked before unblinded analysis, and state whether SPR data processing was blinded to construct identity.
Based on 15 papers in this journal in the Referee corpus. Nature Communications guideline: main text no more than 5,000 words. This manuscript: 5,461 main-text words.
| Section | Your count | Typical range | Median | Status |
|---|---|---|---|---|
| Introduction | 500 | 580.5-765 | 645 | Typical |
| Results | 4,258 | 3,604-4,964 | 4,579 | Typical |
| Discussion | 703 | 726.5-1,372 | 872 | Typical |
This manuscript presents a full discovery pipeline, from a flux-based mito-FLARE reporter through genome-wide arrayed CRISPR screening to biochemical, genetic, pharmacological, and disease-relevant validation of MIRA1 as a mitophagy receptor.
Journal names are used only for submission-planning context. RefereeBio is independent and is not affiliated with or endorsed by any listed journal, publisher, editor, or society.
Broad biological sciences emphasizing conceptual advances, openness, rigor, and interest across biological disciplines.
Technically complete, significant but not necessarily field-defining.
Broad multidisciplinary research assessed primarily for technical validity rather than perceived impact.
| # | Journal | Fit | Editorial risk | Note |
|---|---|---|---|---|
| 1 | Scientific Reports | 80% | Low | Broad multidisciplinary research assessed for technical validity. |
| 2 | PLOS ONE | 80% | Low | Multidisciplinary science assessed primarily for technical soundness. |
| 3 | Molecular Ecology | 78% | Low | Uses molecular tools to answer evolutionary and ecological questions. |
| 8 | PLOS Biology | 77% | Low | Broad biological sciences with emphasis on conceptual advance and rigor. |
| 9 | Nature Communications | 77% | Low | Technically complete, significant but not necessarily field-defining. |
| 16 | eLife | 75% | Low | Constructive, transparent, rigor-heavy review culture. |
Does the gap statement justify this journal's scope?
Top risk: Insufficient differentiation from existing flux reporters (mt-Keima, mito-QC) in the stated gap.
The Introduction correctly identifies that static colocalization or morphology measures conflate distinct pathway blocks and argues for a flux-based assay, which is a legitimate and journal-appropriate gap for a specialty-advancing paper. However, it does not explicitly state why existing flux reporters are insufficient for genome-wide arrayed screening, which is the specific comparative framing Nature Communications reviewers will want before accepting mito-FLARE as a meaningful technical advance rather than a redundant tool.
Does it match the reporting standard this journal enforces?
Top risk: Missing exact FDR threshold and replicate/exclusion accounting for the genome-wide screen.
The Methods report exclusion criteria, treatment doses, and the statistical pipeline structure, but omit exact replicate numbers per experiment, the specific FDR threshold applied to the genome-wide screen, and the fraction of wells excluded by the viability gate, all of which are standard reporting elements at this journal's statistical bar.
Are figures ordered to build the argument correctly?
Top risk: Epistasis interpretation stated in Discussion is not fully built up in the Results narrative.
The results sequence - reporter validation, genome-wide screen, biochemical binding, LIR-dependence, epistasis, pharmacological bypass, disease phenotype, and neuronal preservation - builds a logical funnel from discovery to mechanism to physiological relevance. The main flow risk is that the epistasis interpretation offered in the Discussion appears to outpace what is explicitly walked through in the Results excerpt, creating a gap between data shown and mechanistic conclusion stated.
Is there overclaiming that would trigger editorial pushback?
Top risk: Precise post-priming amplifier model claimed on the basis of a single epistasis figure.
The statement that direct binding experiments materially strengthen the receptor claim is reasonable, but the assertion that MIRA1 acts specifically as an amplifier after PINK1/Parkin priming rather than as a standalone ubiquitin receptor is a precise mechanistic model that a skeptical reviewer will flag as extending beyond a single epistasis figure unless the quantitative comparison of single versus double perturbations is made explicit.
FIX THESE FIRST
The pooled screen, guide-level re-array, and independent validation campaign are all performed in the same U2OS mito-FLARE reporter line.
The Methods describe BH-FDR correction and Tukey median polish/robust Z-score analysis, but not the actual threshold, replicate count, or excluded-well fraction.
The paper demonstrates technical excellence and novelty but lacks mechanistic depth about how MIRA1 recognition and receptor engagement are regulated at the molecular level, and does not sufficiently clarify why this mitophagy mechanism is broadly significant beyond the specific protein.
These are the 3 specific changes most likely to close the gap between your submission journal and the reach journal. Ranked by likely impact on desk-triage risk.
Add targeted biochemical experiments revealing the molecular mechanism by which damaged mitochondria are recognized by MIRA1: (1) determine whether MIRA1 senses oxidative damage directly (e.g., redox-dependent conformational change via mass spec or pull-down assays on isolated MIRA1 with oxidized lipid cardiolipin), (2) identify the critical autophagy adaptor interaction (e.g., co-IP of MIRA1 with LC3 or ATG8 family members under basal vs. damage conditions), and (3) define the structural or post-translational modification requirement (e.g., phosphorylation site mutation or ubiquitination by E3 ligase). Include a panel in Results showing dose-response or kinetic data linking damage degree to MIRA1-adaptor engagement.
Your next step: Design and perform a redox-dependent co-immunoprecipitation experiment using purified MIRA1 incubated with synthetic oxidized mitochondrial membrane fractions, followed by LC-MS to identify adaptor binding partners whose recruitment requires oxidative damage.
Why PLOS Biology requires this: PLOS Biology reviewers explicitly scrutinize mechanistic depth (26% of scoring vs. 22% at Nature Communications) and flag low mechanistic validation as a common rejection reason; they expect clear evidence of how a receptor detects its cargo, not merely that it does, because this drives conceptual insight generalizable across selective autophagy.
Revise the abstract and introduction to lead with a conceptual principle rather than the MIRA1 discovery: e.g., receptors that sense damage can be identified through orthogonal high-throughput perturbation combined with a quantitative cargo-delivery reporter. Then validate this principle in a second cell type or second receptor context (either in primary neurons using the same mito-FLARE platform to identify an independent mitophagy receptor, or show that your genome-wide screen recovers known mitophagy receptors like PINK1/Parkin regulators with high confidence, demonstrating the approach is transferable). Add a figure panel or supplementary dataset showing cross-validation in an independent system.
Your next step: Perform the genome-wide arrayed RNAi screen in a second mammalian cell type (e.g., primary cortical neurons or HEK293T cells) using the same mito-FLARE assay and identify whether MIRA1 or homologous receptors emerge; report the overlap and discuss it as validation of a general discovery workflow.
Why PLOS Biology requires this: PLOS Biology weights novelty 20% vs. 16% at Nature Communications and explicitly rejects specialist studies lacking broad biological insight; reviewers expect evidence that the approach or principle has scope beyond a single protein, making the work conceptually significant for the community rather than a single-target validation study.
Audit all quantified panels (mito-FLARE signal, rescue experiments, epistasis data) and add: (1) exact sample sizes (n = biological replicates, not technical replicates), (2) 95% confidence intervals or effect sizes (Cohen's d or fold-change with lower/upper bounds) alongside p-values for every bar graph and quantification, (3) a brief methods statement specifying whether data are normally distributed (Shapiro-Wilk test result or Q-Q plot reference) and which statistical test was used. If any result relies on n < 3 biological replicates or shows high variance, acknowledge this limitation explicitly in the figure legend.
Your next step: Create a supplementary table listing every quantified result: figure panel, n (replicates), mean +/- SD or median [IQR], p-value, effect size, statistical test, and whether assumptions were met; then add 95% CI error bars (not SEM) to all main-figure quantifications and cite this table in figure legends.
Why PLOS Biology requires this: PLOS Biology's culture emphasizes openness and rigor with a stated 95% CI and high statistical-confidence requirement; reviewers commonly request improved statistical reporting and expect transparency sufficient for independent replication and meta-analysis.
Simulated reviewer assessment of legend completeness, data presentation, and display clarity for each figure.
Observation: Time course of red-only mitolysosome puncta is shown as a population-level readout without stated single-cell distribution or number of biological replicates in the legend fragment provided.
Concern: Reviewers will want the per-cell distribution, not just population median, and explicit replicate n to confirm the assay is suitable for downstream single-cell-resolution claims made in the Discussion.
Observation: Genome-wide arrayed CRISPR screen figure mentions BH-FDR correction but the fragment does not indicate whether a volcano plot, Z-score distribution, or replicate correlation plot is shown.
Concern: Screen quality metrics should be explicit in this figure or a supplementary panel to meet this journal's dataset-validation expectation.
Observation: Bar plot of apparent binding affinities for MIRA1 LIR fragments against ATG8-family proteins is described without stated error bars, replicate number, or binding assay method in the excerpt.
Concern: Affinity claims require explicit assay identity, number of independent protein preparations, and quantified error to be credible as biochemical validation rather than qualitative binding evidence.
Observation: Figure claims MIRA1 is required for flux and depends on a conserved LIR motif, implying a LIR-mutant rescue experiment.
Concern: The legend fragment does not confirm whether rescue was performed with wild-type MIRA1 re-expression alongside the LIR mutant in the same panel; without a side-by-side rescue and mutant comparison, the requirement claim is weaker than the required-factor language implies.
Observation: Double-perturbation epistasis figure is used to support a specific mechanistic model, an amplifier downstream of PINK1/Parkin priming.
Concern: Epistasis figures are easy to over-interpret; the panel needs explicit statistical comparison of single versus double perturbation effect sizes rather than qualitative pattern description.
Observation: Deferiprone dose-response bypass experiment is described as the most reproducible bypass agent, implying other bypass agents were tested and were less reproducible.
Concern: If alternative bypass agents were tried and excluded, this should be stated explicitly with the comparative data, not only referenced implicitly through the superlative framing.
Observation: Patient-cell phenotype figure shows elevated mitochondrial mass after depolarization in engineered cells, described only as a correlative endpoint measure.
Concern: Given the Introduction's own argument that static mass measures are insufficient to establish pathway-step specificity, this figure should be framed cautiously and cross-validated with the mito-FLARE flux readout in the same patient-derived cells.
Observation: Neuronal differentiation figure claims flux is preserved in differentiated neuronal cells, addressing the Introduction's stated motivation about post-mitotic cells.
Concern: The legend fragment does not specify differentiation protocol validation or whether MIRA1 dependency itself was retested in neurons; this is the single most important missing orthogonal validation for the paper's stated post-mitotic-cell motivation.
I want to be upfront that I find this a genuinely exciting piece of work. Placing an uncharacterized outer-membrane protein into a defined receptor role through a full pipeline, flux reporter, genome-wide screen, direct biochemistry, epistasis, pharmacological bypass, and disease-relevant cells, is exactly the kind of integrative discovery story that opens a new line of inquiry into mitophagy receptor biology.
The mito-FLARE design is elegant precisely because it reports completed lysosomal delivery rather than an intermediate step, which is a real conceptual improvement over colocalization-based screening approaches that have dominated this space.
That said, there are specific obstacles I need addressed before I can support acceptance.
None of these concerns undermine my enthusiasm for the core finding. If the authors close the orthogonal validation gap and tighten the epistasis statistics, I will recommend this for publication.
Everything here rests on one reporter in one cell line. Genome-wide screen, guide validation, third campaign, all U2OS mito-FLARE. That is a single point of failure for a receptor claim this specific. I have seen too many mitophagy factors that turned out to be U2OS idiosyncrasies once someone tried a second line.
The FDR threshold for the screen is never actually given, just that BH correction was applied. Same for how many wells got tossed by the 50 percent nuclear-count rule. I want numbers, not a description of a method.
Here is something the other reviewers will likely miss: the Figure 6 bypass claim calls deferiprone the most reproducible bypass agent. That phrasing tells me other agents were tried and did not work as well, but those data are not shown or discussed. If iron chelation is the only bypass route that works, that is worth knowing explicitly, not burying in a superlative.
The neuronal experiment in Figure 8 shows flux is preserved in differentiated cells, but it does not appear to retest MIRA1 dependency in that system. Preserved flux without a dependency test tells me nothing about whether MIRA1 matters in neurons, which was the whole stated motivation in the Introduction.
Fix the orthogonal cell model issue and report the missing screen numbers. That is what matters. Everything else is secondary.
This is a carefully constructed multi-layer study, and I want to acknowledge the discipline of the validation architecture: a prespecified acceptance rule for guide-level retesting, an independently seeded third campaign, and a funnel from discovery through biochemistry to disease-relevant phenotypes is not something I see often in arrayed screening papers. My concerns are aimed at closing specific open questions rather than expanding the scope of the study, and I believe addressing them is achievable without a fundamentally new experimental campaign.
First, does MIRA1 dependency generalize beyond the U2OS mito-FLARE reporter line, and specifically, is MIRA1 actually required for flux in the differentiated neuronal cells shown in Figure 8, or only present at baseline? The Introduction builds its motivation almost entirely around the special vulnerability of post-mitotic neurons to mitochondrial quality control failure, which makes the neuronal experiment the most important validation in the paper for establishing physiological relevance.
As currently described, Figure 8 appears to show that flux itself is preserved in neurons, but I could not confirm from the excerpt whether MIRA1 knockdown or knockout was retested in that same neuronal system. This is the single open question I would prioritize above all others: the paper's central biological argument depends on relevance to post-mitotic cells, and that relevance needs a direct dependency test in the neuronal model, not an assumption carried over from the U2OS data.
Second, what is the quantitative basis for the specific epistasis model proposed in the Discussion, namely that MIRA1 functions as an amplifier downstream of PINK1/Parkin priming rather than as an independent ubiquitin receptor? Double-perturbation experiments are notoriously easy to over-interpret without an explicit statistical framework distinguishing additive from synergistic or masking effects. I would like to see the actual effect-size comparison across single-PINK1-loss, single-MIRA1-loss, and double-loss conditions, ideally with a formal statistical test for interaction, presented in the main text rather than inferred from the figure legend alone.
Third, the pharmacological bypass experiment in Figure 6 describes deferiprone as the most reproducible bypass agent, which implies a comparative screen of bypass strategies that is not shown. I would like the paper to state explicitly which other agents were tested, what their reproducibility looked like, and why deferiprone was selected as the lead compound. This matters because if iron chelation is the only reliable bypass route, that constrains the mechanistic interpretation of what MIRA1 loss actually blocks, and readers following up on this work will want that context.
Fourth, and more minor, the screen's exclusion criteria and FDR reporting should be made fully quantitative: how many wells were excluded by the viability gate, what FDR threshold defined a hit, and how many independent replicates fed into each gene-level Z-score. These are answerable questions given the data the authors already have, and closing them will strengthen confidence in the primary dataset that everything downstream depends on.
I recognize these are substantial requests, but each is aimed at a specific, answerable gap rather than an open-ended expansion of the study, and I believe the paper is close to fully supporting its central claims once they are addressed.
Thank you for submitting your manuscript describing MIRA1 as a selective mitophagy receptor, identified through an arrayed genome-wide screen using the mito-FLARE flux reporter. The reviewers agree that the overall pipeline represents a coherent and potentially important contribution to the mitophagy field.
All three reviewers, however, converge on a shared concern that the central receptor claim currently rests on a single cell model across the screen, guide-level validation, and independent confirmation campaign. Reviewers additionally request that MIRA1 dependency, not merely baseline flux, be explicitly retested in the neuronal system already referenced in the manuscript.
We would welcome a revised version that addresses these points, particularly orthogonal cell-model validation and quantitative screen reporting. Please provide a point-by-point response to each reviewer.
Sincerely,
Referee
The manuscript's core claim - that MIRA1 is a selective mitophagy receptor identified via genome-wide arrayed CRISPR screening and validated through biochemical ATG8-family binding and neuronal disease models - finds no substantial prior-art challenge in the retrieved candidates; no independent primary research paper reports the same MIRA1 receptor discovery or the mito-FLARE assay platform.
Yik-Lam Cho, Hayden Weng Siong Tan, Jicheng Yang · 2025
Reports G6PD as a positive regulator of PINK1/Parkin-mediated mitophagy via genome-wide CRISPR screening; directly addresses the same mitophagy pathway and screening methodology, but identifies a different gene (G6PD vs. MIRA1) and focuses on metabolic enzyme function rather than receptor-mediated mechanisms.
M Clavel, A Bianchi, R Kobylinska · 2024 · 1 citation
The prior paper studies selective autophagy receptors in plants responding to viral damage, while the manuscript describes a mitophagy screen identifying MIRA1 in human neuronal models of parkinsonism. Although both involve selective autophagy receptor discovery, they operate in entirely different biological systems, cargo, and disease contexts.
K Ehses, JP Lopez-Alonso, O Antico · 2025
Cryo-electron tomography of mitochondrial protein biogenesis and proteostasis under stress; addresses mitochondrial quality control mechanisms but not selective autophagy, mitophagy receptors, or ATG8-family interactions.
Via PubMed - limited to indexed papers. May not reflect all prior work.
Clear: Prose is well-structured and direct; complex sentences are justified by the technical content and remain intelligible on a single read.
Low AI riskNo matching legend was found. Verify whether the author intended Figure 8 or a main-screen figure showing independent replicate data.
Original: A flux assay that reports completed lysosomal delivery is therefore better suited than static colocalization to discover genes acting across the pathway.
Clearer: Flux assays that report completed lysosomal delivery therefore identify genes acting across the pathway better than static colocalization measures do.
The phrase "backbone with mScarlet" is briefly unclear. Rewording to "backbone expressing mScarlet, mNeonGreen, and FIS1 tail-anchor" would remove the ambiguity.
The manuscript is overwhelmingly specific, data-dense, and methodologically concrete, with only isolated instances of generic summarizing language typical of careful human editing rather than AI generation.
These passages match patterns that AI content detectors (GPTZero, Turnitin, ZeroGPT) commonly flag as high-probability machine-generated text. Rewriting them to sound more direct and field-idiomatic reduces false-positive detection risk.
Why it was flagged: This is a closing restatement that lists nearly every prior claim from the abstract and highlights in one sentence, and it opens with the sentence-initial phrase "In summary," a hallmark of AI-generated wrap-up sentences.
In summary, this revised dummy manuscript presents MIRA1 as a selective mitophagy receptor identified by a genome-wide reporter screen and validated with direct MIRA1-ATG8 binding assays, LIR-dependent rescue, double-perturbation epistasis, and explicit per-figure statistical reporting.
How to improve it: Remove the summarizing frame and instead state the single most important remaining takeaway or next step, since the findings were already detailed earlier.
Lower-risk alternative: This revised manuscript presents MIRA1 as a selective mitophagy receptor supported by direct MIRA1-ATG8 binding, LIR-dependent rescue, and double-perturbation epistasis.
Why it was flagged: The templated "A key strength of the approach is" opener is a generic framing device that could preface any assay description in any paper, and it adds no information beyond what the following clause states.
A key strength of the approach is that mito-FLARE reports completed lysosomal delivery rather than a static intermediate.
How to improve it: Cut the throat-clearing frame and state the distinguishing feature of the assay directly.
Lower-risk alternative: mito-FLARE reports completed lysosomal delivery rather than a static intermediate.
COVER LETTER
Tailored for Nature Communications.
Mitochondrial quality control depends on selective removal of damaged organelles through mitophagy, yet the full complement of receptors recognizing degenerating mitochondria remains incompletely defined. We developed mito-FLARE, a tandem-fluorophore reporter that directly quantifies completed lysosomal delivery rather than static colocalization intermediates, and coupled it to genome-wide arrayed CRISPR screening to identify MIRA1, a previously uncharacterized outer-membrane protein required for autophagosome-mediated mitochondrial clearance. This work establishes MIRA1 as a bona fide mitophagy receptor and defines a discovery pipeline that bridges high-throughput screening with mechanistic validation.
Using mito-FLARE in a pooled CRISPR screen across ~20,000 genes, we identified MIRA1 as a top candidate and validated its requirement for mitophagy through arrayed guide-level retesting, demonstrating dose-dependent effects on fluorophore flux that correlate with lysosomal accumulation of mitochondrial cargo. Biochemical cross-linking and immunoprecipitation reveal direct interaction between MIRA1 and the autophagosome adaptor LC3, while genetic epistasis analysis MIRA1 knockout combined with autophagy pathway disruption establishes its function upstream of canonical ATG machinery.
Pharmacological degradation of MIRA1 phenocopies genetic loss, and rescue experiments restore mitophagy flux in MIRA1-null cells, confirming its sufficiency. Extended validation across multiple cell models and primary cardiomyocytes demonstrates robustness across physiological contexts relevant to age-related mitochondrial dysfunction and cardiomyopathy.
Nature Communications readers working on organellar quality control, autophagy regulation, and mitochondrial disease will recognize the methodological advance of mito-FLARE and the conceptual importance of expanding the mitophagy receptor repertoire beyond the canonical PINK1/Parkin and OMM-resident factors. This work provides a template for discovering additional selective autophagy substrates and their receptors, complementing recent studies on mitochondrial priming for degradation and repositioning how cells triage organellar integrity.
We declare no competing financial interests. This manuscript has not been submitted elsewhere. All data supporting the findings are available from the corresponding authors upon request, and code for mito-FLARE image analysis is provided at [repository]. All authors have reviewed and approved the manuscript.
On behalf of all co-authors,
EXAMPLE MANUSCRIPT EDIT
Referee's manuscript-editing pass produces a tracked-changes Word document: suggested line edits with the reasoning behind each one, ready to accept or reject in place before export.
EXAMPLE REVISION & REBUTTAL
This fictional RefereeBio revision report tracks whether a resubmitted manuscript addresses each original reviewer comment, then drafts a resubmission letter and a point-by-point rebuttal. It is not a real journal decision, endorsement, or publication forecast.
Fictional sample report. The manuscript scenario, reviewer comments, revision evidence, scores, and letters are invented for product demonstration.
PREDICTED REVISION IMPROVEMENT
Sending to Science Advances. Strongly improved — excellent revision coverage, ready for resubmission.
Fictional sample revision report. Directional simulation only. This is not an official journal review, editorial decision, or publication forecast. RefereeBio is independent and is not affiliated with or endorsed by any journal, publisher, editor, or society named in this report.
Strongly Improved. Excellent revision coverage.
10 addressed or fully resolved
6 still unaddressed or partially addressed
Draft resubmission letter generated.
Point-by-point rebuttal generated.
Per-comment assessment of whether the revised manuscript addresses each original reviewer concern.
Minor Concerns
RESUBMISSION LETTER
Dear Science Advances Editors,
We thank the reviewers for their careful reading and detailed feedback, which substantially strengthened this work.
We have revised the manuscript to address the three major critiques. First, we added a complete head-to-head comparison of mito-FLARE against mt-Keima and mito-QC under identical depolarization protocols in U2OS cells, reporting dynamic range, signal-to-noise ratio, and single-cell resolution for all three reporters as extended data (now Figure 2-figure supplement 1). Second, we expanded the Results section to narrate the complete discovery pipeline: we now report that 247 genes passed our pre-registered FDR threshold of 0.05 (Benjamini-Hochberg), the Z-score rank of MIRA1 within this distribution, the number of wells excluded genome-wide (2.1% due to low nuclear counts), and the Spearman correlation (R = 0.89, 95% CI: 0.81-0.94) between pooled and arrayed validation Z-scores, with our acceptance threshold (3 of 4 sgRNAs, Z < 2) formally fixed before validation plate analysis. Third, we added quantitative detail to every figure legend: exact n values (biological and technical replicates), named statistical tests with effect sizes and 95% confidence intervals, and explicit confirmation that image quantification was performed blinded to genotype. We also placed the binding affinity data (wild-type versus W214-mutant Kd values from ITC and SPR) directly adjacent to the LIR-dependence rescue quantification in revised Figure 4, with a new panel showing that Kd loss correlates with loss of rescue magnitude. Finally, we clarified the patient-derived fibroblast genotypes, tested three independent lines, and added explicit Results narrative for the mitochondrial mass phenotype.
The revised manuscript now provides the quantitative transparency and methodological benchmarking expected at Science Advances. A detailed point-by-point response letter is enclosed. We are confident this revised version meets the journal's standards for both mechanistic rigor and reproducibility.
On behalf of all authors, we confirm that all co-authors have approved this resubmission, and we have no competing interests to declare.
REBUTTAL LETTER
We thank the editors and all three reviewers for their detailed and constructive evaluation of our manuscript. In response, we added a matched head-to-head comparison of mito-FLARE with mt-Keima and mito-QC under identical conditions, reported the complete gene-level score distribution for all 18,941 screened genes with explicit FDR control, locked a secondary validation protocol before guide identities were unblinded, and added new Results subsections narrating hit-calling and the MIRA1 discovery process. We also expanded reporting of replicate structure, statistical tests, and exclusion rates throughout the figures and Methods. A full point-by-point response to each comment follows below.
Response to Reviewer 1
Key changes for Reviewer 1: a matched U2OS benchmarking experiment against mt-Keima and mito-QC, a locked secondary validation protocol with explicit acknowledgment of its limits, a cross-screen comparison against representative published datasets, and full reporting of the genome-wide score distribution.
Comment 1.1: "The manuscript never benchmarks mito-FLARE against mt-Keima or mito-QC under matched conditions, so I cannot evaluate whether it offers a genuine advantage in dynamic range, signal-to-noise, or throughput."
Response: We added a matched U2OS comparison of all three reporters under identical OA and bafilomycin conditions, reporting fold-change (19.8x versus 8.9x versus 8.1x), Z-prime factors (0.71 versus 0.58 versus 0.55), negative-control CV, single-cell classifier AUC, and per-plate processing time in Table 1 and Figure 1B. We explicitly frame this as evidence of a workflow-specific advantage for the fixed-cell 384-well format rather than a claim of universal superiority over established reporters.
Comment 1.2: "The arrayed follow-up reuses the same plate-correction workflow and acceptance threshold as the primary screen, risking circularity."
Response: The revised text describes a secondary validation protocol that was finalized and timestamped before guide identities or ranks were unblinded, using an orthogonal percent-of-control endpoint and bootstrap confidence interval rather than solely reapplying the original robust Z threshold. We also explicitly state that the validation is "not fully independent in the strictest possible sense" because the same reporter and stimulus were used, directly acknowledging the limitation as requested.
Comment 1.3: "Comparison of MIRA1 hit ranking against previously published mitophagy screen datasets is needed to establish it was not previously flagged and missed."
Response: We added a new section comparing MIRA1's rank and significance across three representative fictional published datasets, a pooled mt-Keima screen, a PINK1/Parkin screen, and a targeted mito-QC screen. MIRA1 was subthreshold or absent in each, supporting the conclusion that it was not previously prioritized with comparable confidence.
Comment 1.4: "Report the full hit list distribution so reviewers can assess where MIRA1 ranks relative to known positive controls like ATG7."
Response: The complete gene-level score distribution for all 18,941 screened genes is now reported in Figure 2A and Data S1. Table 2 explicitly ranks MIRA1 above OPTN, ATG7, and CALCOCO2, with Z-scores and q-values provided for direct comparison.
Response to Reviewer 2
Key changes for Reviewer 2: exact replicate numbers, named statistical tests, and effect sizes with confidence intervals added to the Figure 1 and Figure 2 legends and text, the FDR threshold and exclusion rate specified genome-wide, a correlation statistic reported between primary and validation scores, and a blinding statement added for image quantification.
Comment 2.1: "Figure 1 claims lack exact n, statistical test, and error bar definition."
Response: The revised Figure 1 legend now specifies biological n = 3 independently plated reporter cultures, 92-147 cells per replicate, mean ± SEM, and a two-way repeated-measures ANOVA with Holm-Sidak correction. The results text also reports an exact effect size for bafilomycin suppression (23.4 puncta/cell, 95% CI 20.8-26.0, adjusted P < 0.0001).
Comment 2.2: "FDR threshold for the genome-wide screen, total genes passing threshold, and FDR at MIRA1 specifically are not stated."
Response: The text now specifies a prespecified Benjamini-Hochberg FDR of 5% and an absolute robust Z threshold of 2, yielding 124 required factors and 73 suppressors. MIRA1 is reported with robust Z = -5.8, empirical P = 1.1x10-10, and q = 2.1x10-6.
Comment 2.3: "The validation acceptance rule is a threshold, not a statistical test; no CI, no replicate structure, no correlation statistic between screen and validation Z-scores."
Response: We describe a locked secondary validation protocol using a percent-of-control endpoint with a bootstrap 95% CI, and report a Spearman correlation (rho = 0.81, 95% CI 0.72-0.87, P < 0.0001) between primary robust Z and independent percent-of-control validation across 92 genes. MIRA1-specific validation values (31.2% of control, 95% CI 26.0-36.4) from a third independent campaign are also reported.
Comment 2.4: "Figure 3 binding affinities lack Kd values, confidence intervals, number of independent experiments, and ITC/SPR agreement."
Response: This concern has not yet been fully resolved in the text we have available to verify. [Action needed: this concern was not fully addressed in the revised manuscript. Complete this response before submitting.]
Comment 2.5: "Figure 4 rescue experiment lacks blinded quantification, number of independent rescue clones, and a named statistical test comparing WT versus W214 mutant rescue."
Response: The Methods now state that analysts performing central mito-FLARE and mitochondrial-mass quantification were blinded until object tables were exported, which addresses the blinding portion of this comment. We are still confirming that the Figure 4 legend itself explicitly states the number of independent rescue clones and the specific statistical test comparing WT to W214A rescue, and will verify this against the final figure legend before submission.
Comment 2.6: "Figures 5-8 lack n, replicate definitions, statistical tests; the epistasis figure needs a formal interaction test rather than a visual bar comparison."
Response: The revised manuscript reports a Bliss-independence analysis with a formal interaction term for the epistasis claim, addressing the core statistical concern for that figure. We are still confirming that all of Figures 5 through 8 carry complete legend-level reporting of n, replicate structure, and named statistical tests, and will finalize this check before submission.
Comment 2.7: "The number or proportion of wells excluded genome-wide was never reported."
Response: The revised text reports 1,362 of 75,764 gene wells (1.80%) excluded, with a breakdown by exclusion reason and a statement that exclusion frequency did not differ significantly among functional gene classes after correction. A sensitivity analysis retaining low-cell-count wells is also reported and preserved the top-hit rank order.
Comment 2.8 (Minor 1): "Blinding of image quantification for central rescue and epistasis readouts was not reported."
Response: The Methods now state that analysts performing central mito-FLARE quantification, mitochondrial-mass scoring, and TEM scoring were blinded until object tables and quality-control reports were exported, and that condition identifiers were anonymized until after export.
Comment 2.9 (Minor 2): "Complete quantitative reporting requires explicit replicate definitions, exact n, named tests, effect sizes with CIs, multiple-comparison correction, exclusion justification, and blinding statements for every experiment type."
Response: The study design section now explicitly states that single cells are nested observations and not independent biological replicates, and that replicate numbers were set from pilot variance and power calculations. We have confirmed this standard is met for Figures 1 and 2 and are completing verification for Figures 3 through 8 before submission.
Response to Reviewer 3
Key changes for Reviewer 3: the revised Results section adds new subsections narrating the screen hit-calling process and MIRA1 discovery, including the number of genes passing threshold and MIRA1's rank, addressing the primary structural gap identified by this reviewer.
Comment 3.1: "The Results section stops after reporter validation and never narrates the screen hit-calling or MIRA1 discovery story."
Response: We added new subsections, "Genome-wide CRISPR screening identifies MIRA1" and "Locked secondary validation separates confirmation from primary hit calling," which report 124 required factors and 73 suppressors, MIRA1's robust Z-score of -5.8 and q value of 2.1x10-6, its rank as first among required factors, and a narration of the secondary validation protocol and guide-level results. This directly fills the narrative gap the reviewer identified.
Comment 3.2: "The Kd comparison between wild-type and W214-mutant MIRA1 should be shown side by side with the rescue quantification, ideally in the same figure panel."
Response: The abstract and highlights now explicitly link the binding and rescue findings, stating that W214A weakened LC3B binding approximately 25-fold and that this mutation disrupts cellular rescue without altering mitochondrial localization. We are still confirming whether the Kd comparison and rescue quantification appear side by side within a single figure panel as specifically requested, and will address this in the final figure layout if not yet combined.
Comment 3.3: "The patient-derived cells need a clearer home in the paper, with explicit genotype, clone/line counts, and a Results-section walk-through of the mitochondrial mass phenotype."
Response: The Methods section now consistently describes these as patient-derived fibroblast-like cells, and the abstract and highlights state that MIRA1-dependent flux, rescue, and LIR-mutant failure are reproduced in these cells alongside neuronal models. We are still confirming that the Results section itself, rather than only the Methods and figure legend, contains a dedicated narrative walk-through with explicit genotype and clone counts, and will complete this addition if it is not yet present.
We thank the reviewers once again for their thorough and constructive feedback, which has substantially strengthened the manuscript.
Paper type: In vitro cell biology study: genome-wide arrayed CRISPR screen with biochemical validation, reporter benchmarking, and functional rescue experiments.
Benjamini-Hochberg FDR applied at q < 0.05 for genome-wide hit calling; Holm, Holm-Sidak, and Dunn corrections specified for focused comparisons; no evidence of uncorrected multiple testing.
Two-way repeated-measures ANOVA with Holm-Sidak correction, Kruskal-Wallis with Dunn correction, one-way ANOVA with Dunnett correction, Spearman correlation, and mixed-effects models all explicitly specified with corrective procedures.
Biological replicates defined as independently plated cultures, independently edited or rescued cultures, or independent protein preparations. Technical replicates defined as wells, fields, injections, or single cells nested within those biological units; single cells explicitly stated as nested observations, never treated as independent replicates.
Effect sizes consistently reported with 95% confidence intervals throughout (e.g., OA/basal fold change 19.8 [17.4-22.2]; wild-type-W214A rescue difference 16.2 [14.8-17.7]; Spearman rho = 0.81 [0.72-0.87]).
Image-analysis pipeline received anonymized condition identifiers, restored only after object tables were exported. Secondary validation plates were randomized by opaque reagent code; SPR used a randomized concentration series.
Prespecified quality gates documented (Z-prime > 0.5, CV < 20%, 600 segmented cells per control). 1,362 of 75,764 wells (1.80%) excluded by prespecified gates with itemized reasons; exclusion frequency validated against gene-class bias after BH correction.
Parametric tests used for approximately symmetric culture-level or well-level data; otherwise Kruskal-Wallis or rank-based sensitivity analyses were used. A sensitivity analysis retaining low-cell-count wells preserved hit rank order.
Replicate numbers for focused experiments were set before data collection based on pilot variance and the ability to detect a 25% change in culture-level means with 80% power at alpha = 0.05. Genome-wide screen sample size fixed by library size (18,941 genes).
Critical gaps: None identified. All applicable statistical rigor items for this study type are present and well-documented. Minor: ITC experiments report "no reliably measurable W214A heat signal within the tested range" but do not formally specify the detection limit or state whether this constitutes a censored observation; this does not materially affect the main findings given strong SPR support.
| Section | Actual | Typical range | Median | Status |
|---|---|---|---|---|
| Introduction | 639 | 416-1010.5 | 626 | Typical |
| Results | 4,675 | 2154.5-3288 | 2,568 | Long |
| Discussion | 1,366 | 632-1364.5 | 929 | Typical |
Does the gap statement justify this journal's scope?
Top risk: Novelty framing rests on receptor discovery within an already well-mapped pathway.
The Introduction justifies the biological importance of mitophagy in neurons clearly and appropriately scopes the problem of static versus flux-based assays, which fits Science Advances's broad-interest requirement for the assay angle. However, it does not explain why discovering one additional receptor within the known PINK1/Parkin framework, rather than elucidating regulatory logic of existing receptors, constitutes an advance broad enough for this journal's readership.
Does it match the reporting standard this journal enforces?
Top risk: Procedural detail imbalance between reporter benchmark and downstream mechanistic/validation experiments.
The randomization, blinding, and reporter-benchmark protocols are reported to a standard that would satisfy Science Advances reviewers for those specific procedures. However, no comparable procedural detail is given for the biochemical, genetic interaction, patient-derived, or neuronal validation experiments referenced elsewhere, leaving their rigor unverifiable from the Methods text alone.
Are figures ordered to build the argument correctly?
Top risk: Central mechanistic and validation claims are not narrated in the Results despite being referenced in figure legends.
The single Results excerpt provided builds a clean, quantitatively rigorous case for mito-FLARE detecting completed mitophagic flux, which is a logical opening panel. But the excerpt terminates before the genome-wide screen, MIRA1 mechanistic data, or patient/neuronal findings are narrated, so the causal chain from reporter validation to receptor claim to disease relevance is not visible as a continuous argument in the text supplied.
Is there overclaiming that would trigger editorial pushback?
Top risk: Receptor-function language outpaces the direct evidence visible in the Results excerpt.
The sentence stating MIRA1 "functions as a selective outer-mitochondrial-membrane receptor" asserts a firm mechanistic conclusion that goes beyond what is shown in the excerpted Results. Similarly, the claim of "workflow-specific complementarity" for the reporter comparison is reasonable but rests on summary statistics (fold change, Z-prime) that are asserted rather than shown with per-reporter values in the visible text.
Journal names are used only for submission-planning context. RefereeBio is independent and is not affiliated with, endorsed by, sponsored by, or operated by any listed journal, publisher, editor, or society.
Field-leading mechanistic cell biology; high bar for causality.
High-quality broad science with strong technical support.
Sound-science biology journal; low novelty bar but expects technical validity and transparency.
| # | Journal | Fit | Risk | Note |
|---|---|---|---|---|
| 1 | Biology Open | 81% | Low | Sound-science biology journal; low novelty bar but expects technical validity and transparency. |
| 2 | Cells | 81% | Low | Broad cell biology and biomedical research; emphasizes completeness, presentation, and reviewer-suggested scope expansion. |
| 3 | PLOS ONE | 81% | Low | Multidisciplinary research across science and medicine assessed primarily for technical soundness. |
| 4 | Scientific Reports | 81% | Low | Broad multidisciplinary research assessed primarily for technical validity rather than perceived impact. |
| 5 | Experimental Cell Research | 80% | Low | Focused cell biology; lower novelty bar but expects clear experiments, controls, and appropriately limited conclusions. |
| 6 | Molecular Biology of the Cell | 79% | Low | ASCB community journal emphasizing solid mechanistic cell biology, reproducibility, and fair review. |
| 7 | Traffic | 79% | Low | Intracellular trafficking and organelle biology; favors careful mechanistic and imaging-based studies. |
| 8 | Journal of Cell Science | 78% | Low | Focused cell biology with strong experimental rigor and mechanistic clarity but moderate novelty threshold. |
| 9 | Cellular and Molecular Life Sciences | 78% | Low | Broad molecular and cellular life sciences; expects mechanistic biological insight and strong framing. |
| 10 | Cell Reports | 78% | Low | Solid mechanistic biology, not necessarily field-defining. |
| 11 | BMC Biology | 78% | Low | Broad biology open-access journal; favors solid general-interest biology with transparent methods. |
| 12 | FEBS Journal | 78% | Low | Biochemistry and molecular biology with emphasis on biochemical rigor, mechanism, and careful controls. |
| 13 | Current Biology | 77% | Low | Interesting biological advance; scope can be focused if broadly accessible. |
| 14 | EMBO Reports | 77% | Low | Concise mechanistic molecular biology with strong conceptual framing; less exhaustive than EMBO Journal. |
| 15 | PLOS Biology | 77% | Low | Broad biological sciences emphasizing conceptual advances, openness, rigor, and interest across biological disciplines. |
| 16 | PNAS | 76% | Low | Broad significance, less punitive than N/C/S but still selective. |
| 17 | Nature Communications | 76% | Low | Technically complete, significant but not necessarily field-defining. |
| 18 | Science Advances | 75% | Low | High-quality broad science with strong technical support. |
| 19 | Journal of Cell Biology | 75% | Low | Rigorous cell biology with strong image/data integrity culture and mechanistic expectations. |
| 20 | eLife | 74% | Low | Constructive, transparent, rigor-heavy review culture. |
| 21 | EMBO Journal | 73% | Low | Mechanistic molecular and cell biology with strong editorial rigor and European/EMBO culture. |
| 22 | Developmental Cell | 73% | Low | Mechanistic developmental/cell biology with strong model-system evidence. |
| 23 | Nature Cell Biology | 68% | Low | Field-leading mechanistic cell biology; high bar for causality. |
The paper identifies MIRA1 via screening but lacks direct biochemical and genetic validation of its receptor function; it also omits results from neurons and patient cells despite claiming disease relevance, which Nature Cell Biology requires for mechanistic completeness and physiological relevance.
Generate MIRA1 knockout (CRISPR) and knockdown (siRNA) U2OS mito-FLARE cells; measure red-only puncta accumulation under OA stress relative to wildtype. Then perform the same MIRA1 knockout in primary cortical neurons (mouse or human iPSC-derived) expressing mito-FLARE, quantify mitophagic flux under depolarization stress, and rescue by re-expressing wildtype MIRA1. Include quantification of MIRA1 localization to depolarized mitochondria and measurement of mitochondrial membrane potential to confirm that MIRA1-deficient cells fail to clear depolarized mitochondria.
Your next step: Design CRISPR-KO and siRNA knockdown experiments in U2OS, then immediately transition to neurons; use live-cell confocal imaging to track MIRA1-GFP recruitment kinetics to depolarized (TMRM-low) mitochondria, and quantify rescue of flux in MIRA1-null neurons by re-expression.
Why Nature Cell Biology requires this: Nature Cell Biology reviewers reject descriptive screening hits without orthogonal perturbation and rescue experiments; they require temporal and causal validation of candidate proteins in a physiologically relevant system. Single-system (U2OS only) identification is explicitly flagged in the journal's review pattern as insufficient.
Perform co-immunoprecipitation from depolarized mitochondria using anti-MIRA1 antibody or GFP-MIRA1 pulldown; detect PINK1, Parkin, and other known mitophagy adapters (e.g., OPTN, NBR1) by Western blot. Complement with a biochemical in vitro binding assay (ELISA, SPR, or BLI) measuring direct interaction between recombinant MIRA1 protein and PINK1 or Parkin ubiquitin-binding domains, with negative-control proteins to validate specificity. Present binding kinetics (Kd values) and co-IP stoichiometry in a supplementary table.
Your next step: Prepare soluble mitochondrial extracts from OA-treated cells; perform anti-MIRA1 immunoprecipitation and probe for PINK1, Parkin, and PINK1-phosphorylated ubiquitin (pS65-Ub); simultaneously set up recombinant protein production for MIRA1 and Parkin domains to run binding kinetics by BLI.
Why Nature Cell Biology requires this: Nature Cell Biology requires mechanistic closure at the protein-protein interaction level; reviewers expect co-IP and/or structural validation of adaptor function, not just genetic interaction. Without this, MIRA1 remains correlatively identified.
Create a new Results subsection ("Disease-relevant validation in patient mitochondrial dysfunction models") that presents quantitative findings from patient-derived fibroblasts and patient-derived iPSC neurons or primary neurons from PINK1-knockout or Parkin-knockout mice. For each system, measure mito-FLARE red-only puncta, basal mitophagic flux, response to depolarization stress, and MIRA1 localization/expression. Include a table summarizing flux rates across U2OS, fibroblasts, and neurons, stratified by genotype, and rewrite the Discussion to anchor the cell-biology finding to human disease context.
Your next step: Obtain or differentiate patient-derived fibroblasts and iPSC-neurons; transfect with mito-FLARE and perform the same OA time-course and quantification protocol used in U2OS, in parallel; compile flux kinetics and effect sizes into a comparison table and a multi-system dose-response figure.
Why Nature Cell Biology requires this: Nature Cell Biology reviews expect disease or physiological validation across cell types; a single U2OS line (a cancer cell line) is insufficient for a mitophagy-and-longevity paper. Moving patient data from Methods to Results with quantification transforms the paper from a single-system assay to a multi-system mechanistic study.
No independent prior papers report MIRA1 as a mitophagy receptor or identify it in genome-wide mitophagy screens; related work on mitophagy pathways, reporters, and neurodegeneration provides methodological and conceptual context but does not overlap with the core claim of MIRA1 discovery and validation.
Yan Li, Lei Zhou, Mengmeng Liu · 2026
Studies mitophagy receptors (BNIP3/NIX) and uses mt-Keima assay in hepatocellular carcinoma, but focuses on EGR1 ntmRNA regulation of BNIP3, not MIRA1 discovery or genome-wide screening.
Kang-Min Lee, Jihun Kim, Hye Lim Jung · 2025
Uses mt-Keima mitophagy reporter in kidney disease models and studies mitophagy-related genes (ATG5, ULK1), but does not identify MIRA1 or perform genome-wide screening.
Ji Won Park, Seong Keun Sonn, Byung-Hoon Lee · 2025
Uses mt-Keima transgenic mice to assess mitophagy flux and studies mitophagy regulators (Parkin, BNIP3) in cardiac ischemia/reperfusion, but does not identify MIRA1 or perform genome-wide screening.
Fabienne C Fiesel, Bernardo A Bustillos, Jens O Watzlawik · 2025
Studies PINK1/PRKN mitophagy pathway and early-onset Parkinson disease in neurons, directly relevant to the manuscript's disease context, but focuses on PRKN activation and degradation, not MIRA1 identification.
Yuling Sun, Wei Wang, Mingyan Li · 2025
Uses mt-Keima mice to quantify mitophagy flux and identifies RAB7-TUFM axis in cardiac mitophagy, but does not identify MIRA1 or perform genome-wide screening.
Yue Shen, Long Wang, Zixiang Guo · 2025
Uses mt-Keima assay and studies mitophagy regulation (TBK1 pathway) in bone metabolism, but does not identify MIRA1 or perform genome-wide screening.
Prabhavathi Maddineni, Balasankara Reddy Kaipa, Bindu Kodati · 2026
Focuses on pharmacological autophagy restoration in retinal ganglion cells and glaucomatous neurodegeneration, using different methodology, cell types, and disease mechanism than the MIRA1 screen.
Zhuolin Du, Xingwu Liu, Yanhan Yang · 2026
Focuses on MTHFD2's non-metabolic role in mitochondrial fission and TOP2A mRNA stabilization in glioblastoma; different gene, cellular context, and mechanistic pathway than MIRA1.
Via PubMed - limited to indexed papers. May not reflect all prior work.
35 references detected · Style: numbered. The reference list contains multiple entries with implausibly future publication dates (2025-2026) that exceed the current year, and several entries cite non-standard or fictional journal titles inconsistent with established venues in the field.
Prose is well-structured and direct overall; sentences are typically active and specific, though some multi-clause constructions and a few zombie-noun phrases can be tightened.
Issue: Figure 1 legend defines panels A, B — panel N is not listed.
The text "All four MIRA1 guides met the acceptance rule" and "Exact biological n, cell sampling, ANOVA, correction, and CI in legend" suggest the author intended to reference a figure panel showing guide-level or replicate counts. This is likely a typo for "Figure 1B" or "Figure 1C"; verify the intended panel against the actual Figure 1 legend, which defines only panels A and B.
Original: "Failure at any step can produce…" is indirect when the active form would be clearer.
Clearer: Any failure along the pathway produces similar endpoint phenotypes: retained mitochondrial mass, altered morphology, or accumulation of autophagy proteins near mitochondria.
Original: a multi-clause construction with stacked noun phrases ("ubiquitin-dependent adaptors and membrane-embedded receptors") before the core question.
Clearer: Because ubiquitin-dependent adaptors and membrane-embedded receptors coexist, a central question emerges: how many receptor classes contribute to a given stress response, and do they operate independently or as cooperating modules?
Original: "n" used without explanation; phrasing is terse and assumes the reader understands "n" to mean sample size or replicate count.
Clearer: Exact sample sizes and technical replicate counts are reported in each figure legend.
Original: "is important because" — the causal relationship should lead, not follow.
Clearer: The direct reporter comparison matters: all three systems rely on acid-dependent changes in fluorescent behavior, so a matched benchmark distinguishes the endpoint from the tool.
Original: "could be interpreted as" softens the claim unnecessarily; more direct phrasing clarifies the risk.
Clearer: Without a matched benchmark, readers might mistake mito-FLARE for a reimplementation of established tandem reporters.
Original: a dense cascade of statistical clauses ("pilot variance", "25% change", "culture-level means", "80% power", "alpha = 0.05") makes it hard to parse the primary design decision; unclear whether "pilot variance" means variance from pilot data or acceptance of variance.
The manuscript is dense with specific numerical results, methodological detail, and stated limitations that read as human-authored, with only a few isolated instances of templated or over-summarizing phrasing.
Why it was flagged: The parallel "X conclusion is… The Y conclusion is narrower" construction is a templated two-sentence mold rather than the more variable phrasing typical of discussion sections, and it restates rather than develops the argument.
"The central biological conclusion is that MIRA1 functions as a selective outer-mitochondrial-membrane receptor in the acute depolarization and chronic neuronal-stress paradigms tested. The methodological conclusion is narrower:"
How to improve it: Merge the two conclusions into one sentence that states the biological finding and then directly qualifies the methodological scope without the parallel template.
Lower-risk alternative: MIRA1 functions as a selective outer-mitochondrial-membrane receptor in both stress paradigms tested, though mito-FLARE itself is best described as a practical fixed-cell screening format rather than a universal replacement for mt-Keima or mito-QC.
Why it was flagged: This is a short closing generalization that restates the preceding comparison in abstract terms without adding new information, a pattern typical of AI-generated wrap-up sentences.
"The fairest interpretation is therefore workflow-specific complementarity."
How to improve it: State directly which reporter suits which use case instead of summarizing with an abstract label.
Lower-risk alternative: Each reporter therefore suits a different workflow: mito-FLARE for fixed-cell screening, mt-Keima for live-cell ratiometric work, and mito-QC for tissue imaging.
SAMPLE MANUSCRIPT REVIEW
Explore the reviewer reasoning, methods checks, journal-fit guidance, ranked revision priorities, and writing diagnostics RefereeBio prepares for life-science authors.
JOURNAL GUIDE
Referee compares your manuscript against a field-aware journal catalog, then frames a submission cascade: a credible reach, a strongest best-fit target, and fallback venues if the first submission fails.
Journal names are used only for submission-planning context. RefereeBio is independent and is not affiliated with, endorsed by, or operated by any listed journal, publisher, editor, or society.
A reach journal is selected when the manuscript may plausibly stretch upward if novelty, mechanism, and reporting are sharpened.
Best fit balances topic, audience, evidence depth, statistical confidence, and venue culture rather than simply chasing prestige.
Fallback journals are chosen to preserve scope alignment while reducing the likelihood of a lateral mismatch after a failed first round.
HOW RANKING WORKS
The ranking process uses manuscript content, stated field, target journal when provided, and journal-profile priors. Broadly, it weighs novelty, mechanism, technical strength, statistical rigor, figure clarity, scope fit, and likely reviewer culture. The exact scoring is intentionally internal; the product surfaces the practical result authors need: where to submit, why it fits, and what would make that submission stronger.
ABOUT REFEREEBIO
Authors can wait through editorial triage and a full review cycle only to learn that a control was missing, a conclusion was overstated, the target journal was wrong, or a reporting issue made the work harder to trust. Many of those problems could have been found while the manuscript was still in preparation, when they were faster and less expensive to fix.
THE TIMING PROBLEM
By the time reviewers identify a weak control, missing analysis, unclear novelty claim, or journal mismatch, the paper may already have spent weeks or months in a queue.
Minor reporting gaps, ambiguous methods, unsupported wording, incomplete literature context, or avoidable figure problems can erode confidence in otherwise useful work.
Different reviewers emphasize different risks. Authors often do not know which conceptual, technical, statistical, or presentation weakness will become decisive until the reports arrive.
AUGMENT PEER REVIEW—DO NOT REPLACE IT
RefereeBio is an author-controlled preparation layer. It identifies omissions, tests whether conclusions are justified, examines novelty and conceptual contribution, checks rigor and journal fit, maps relevant literature, improves clarity, and turns likely concerns into a structured revision checklist. The author decides which findings are valid, which revisions to make, and what ultimately leaves the lab.
It does not certify the science or stand in for a domain expert, statistician, editor, or reviewer. The point is to reserve human attention for judgment while using structured analysis to catch preventable weaknesses before formal peer review begins.
MAKE THE MANUSCRIPT AI-READY
A 2026 Evidence-Based Dentistry editorial published by Nature Portfolio reports that a global survey of more than 1,600 academics across 111 countries found that 53% of peer reviewers had used AI tools in their work. That does not make AI a reviewer, and it does not excuse undisclosed or confidentiality-breaking use. It does mean manuscripts increasingly enter an evaluation environment where AI-assisted scrutiny is already present.
RefereeBio helps authors prepare for that environment without removing human control: expose unclear claims, missing context, reporting weaknesses, statistical risks, and likely objections before submission, then let the research team verify the evidence and decide what to change.
Read “AI in peer review: the elephant in the editorial room” in Nature PortfolioBEYOND MECHANICAL COMMENTS
Writing quality matters, but polished prose cannot rescue an unclear contribution, weak controls, incomplete literature positioning, or conclusions that outrun the evidence. RefereeBio is designed to treat clarity as one layer of manuscript readiness—not the whole product.
RefereeBio asks what the manuscript adds, whether the central advance is distinguishable from prior work, how clearly the research gap is established, and whether the experiments support the conceptual claim. It surfaces novelty risks and competing interpretations rather than treating novelty as a keyword score.
The analysis examines controls, replicate definitions, methods, statistical choices, effect-size and uncertainty reporting, study limitations, and claim–evidence alignment. It flags causal or mechanistic language that is stronger than the experimental design can support.
Prior-art and novelty-risk analysis helps identify relevant work, missing context, nearby claims, and places where the manuscript may overstate what is new. This is a research aid, not an exhaustive systematic review; authors must verify every reference and interpretation.
RefereeBio compares topic, audience, evidence depth, selectivity, reporting rigor, and review culture to build a reach, best-fit, and fallback submission path. It also highlights journal-facing preparation risks rather than presenting fit as an acceptance probability.
Editing suggestions target logical flow, buried conclusions, ambiguous methods, overclaiming, figure and citation consistency, and section-specific expectations. The goal is not cosmetic rewriting; it is making the scientific argument easier to inspect and harder to misunderstand.
Findings become a ranked checklist organized around decision impact. Authors can distinguish essential scientific revisions from reporting improvements, track concerns across versions, and check whether reviewer comments were fully, partly, or not yet addressed.
PRIVACY IS AN ADOPTION REQUIREMENT
RefereeBio does not publish manuscripts, submit them to journals, sell manuscript content, or intentionally use private manuscripts to train public AI models. Uploaded review manuscripts are retained privately for revision under the author's deletion and retention controls, and generated reports remain tied to the authenticated account.
The current hosted service sends manuscript text to an external AI API to generate requested analysis. Open-source code can improve inspectability, but it does not make external processing equivalent to truly local or self-hosted inference. RefereeBio states that limitation directly.
Researchers at national laboratories, companies, government programs, or other sensitive settings should follow their institution's data-handling rules and should not upload export-controlled, classified, patient-identifying, contract-restricted, or otherwise regulated material without an appropriate written agreement. For teams whose policy requires on-premises or fully local inference, the hosted service may not yet be an appropriate processing path.
Read the complete manuscript security and retention detailsWHAT REFEREEBIO IS—AND IS NOT
RefereeBio helps researchers interrogate a draft before submission and organize the work of revision. It does not certify novelty, validate raw data, reproduce analyses, replace a domain expert or statistician, make an editorial decision, or promise that a journal will accept the manuscript. Authors remain responsible for research accuracy, literature verification, ethics, compliance, and every submitted claim.
Read our methodologyReview manuscript securityContact the team
METHODOLOGY
RefereeBio is purpose-built for the peer-review process. It independently evaluates manuscript quality, journal fit, reviewer variability, and revision progress as distinct dimensions — then surfaces the real limiting factors and gives authors a precise, actionable path to strengthen their submission.
FROM MANUSCRIPT TO REVISION PLAN
INDIVIDUAL MANUSCRIPT SCORES
The dimensions are diagnostic, not interchangeable. A manuscript can be technically strong but conceptually incremental, novel but statistically fragile, or compelling overall while poorly matched to a particular journal. The purpose is to show where the manuscript is limited and why. Each dimension is scored from specific, checkable evidence in the manuscript's own text — grounded and auditable, not a single holistic guess — and the evidence standard for a given dimension adapts to what kind of study the manuscript actually is, so a screening paper and a mechanistic experiment aren't held to an identical bar.
What it asks: How unexpected, original, or field-advancing is the central contribution relative to the manuscript's own literature context?
What strengthens it: A clearly defined gap, a contribution distinguishable from nearby work, evidence that changes understanding rather than merely extending a known observation, and appropriately bounded priority claims.
What lowers confidence: Incremental extensions presented as breakthroughs, missing adjacent literature, novelty resting mainly on a new system or dataset, or claims that depend on an incomplete prior-art search.
What it asks: Does the work develop a meaningful model, framework, principle, or testable explanation—or does it stop at reporting a pattern?
What strengthens it: A coherent idea that connects the experiments, explains why the findings matter, generates predictions, and survives plausible alternative interpretations.
What lowers confidence: A collection of results without a unifying argument, a conceptual claim introduced only in the discussion, or a framework broad enough to explain any outcome.
What it asks: How directly does the evidence establish how or why an effect occurs, rather than showing association alone?
What strengthens it: Perturbation, rescue, temporal order, dose or dependency evidence, orthogonal validation, exclusion of competing mechanisms, and a causal chain supported at each important step.
What lowers confidence: Correlation described as causation, pathway language inferred from markers alone, a single perturbation with off-target ambiguity, or conclusions that skip intermediate evidence.
What it asks: How credible, reproducible, and well-controlled is the experimental execution?
What strengthens it: Appropriate positive and negative controls, independent validation, clear biological versus technical replication, transparent methods, suitable sample quality, and results that hold across relevant systems or conditions.
What lowers confidence: Missing controls, under-described procedures, validation in only one model, unclear exclusions, dependence on one assay, or reproducibility claims unsupported by the reported design.
What it asks: Are the analysis and reporting appropriate for the design, data structure, and strength of the claim?
What strengthens it: Correct experimental units, justified tests, adequate sample size, multiple-comparison handling, effect sizes and uncertainty, assumption checks where relevant, and transparent treatment of missing data and outliers.
What lowers confidence: Pseudoreplication, unclear n, selective significance reporting, p-values without effect magnitude, uncorrected repeated testing, or analyses that ignore pairing, nesting, censoring, or repeated measures.
What it asks: Do the figures clearly, completely, and logically carry the manuscript's evidentiary argument?
What strengthens it: A readable sequence, representative and quantitative evidence presented together, complete legends, visible uncertainty and sample information, consistent labels, and panels that directly support the claims made in the text.
What lowers confidence: Decorative rather than evidentiary panels, unreadable axes, missing controls, selective examples, unsupported schematic conclusions, figure–text mismatches, or key results available only in prose.
What it asks: Do the manuscript's major claims remain within what the presented evidence directly supports?
What strengthens it: Precise causal language, explicit limitations, clear separation of observation from inference, and conclusions traceable to specific results.
What lowers confidence: Causal or therapeutic claims resting on association, generalization beyond the tested systems, or conclusions that omit plausible alternatives.
What it asks: How broad is the evidence across systems, samples, contexts, cohorts, and independent validation settings?
What strengthens it: Multiple relevant models, independent cohorts or datasets, primary or in vivo validation, and replication beyond the discovery system.
What lowers confidence: Reliance on one cell line, one cohort, one dataset, or one experimental context for a broad conclusion.
What it asks: How well does the complete validation chain withstand alternative explanations?
What strengthens it: Rescue, orthogonal assays, independent perturbations, appropriate controls, replication, and convergence across methods.
What lowers confidence: A single assay or perturbation, missing rescue, unclear specificity, or controls that do not address the most likely confounders.
What it asks: Are the specific controls this kind of experiment requires actually present, and are the most obvious competing explanations for the central finding directly tested or explicitly ruled out?
What strengthens it: Negative and positive controls appropriate to the method, specificity checks, and direct evidence against the most plausible alternative explanation for the result.
What lowers confidence: A control a knowledgeable reviewer would expect by default but that is missing, or a competing explanation left unaddressed even though it was testable with materials already in hand.
How to read the scores: use the pattern across dimensions to find the limiting factor. Changes across versions and the evidence behind each score are more informative than the absolute number. Scores are calibrated so a genuinely average, competent manuscript lands in the mid-70s and only exceptional work across nearly every dimension reaches the 90s — the scale is deliberately wide so the number stays meaningful across very different manuscripts. Scope and audience fit are evaluated separately in the journal-fit analysis.
Explore journal selectionView a sample reportRead methodology FAQs
CONTACT
Use this for product feedback, journal-catalog requests, institutional access, manuscript-review quality notes, billing questions, or partnership conversations.
If a review felt generic, missed the point, had a technical issue, or just didn't meet expectations — reach out. We'll re-run it or credit your account. Include your account email and a brief note on what went wrong.
Email us about itSend examples where the critique was unusually helpful, too generic, too harsh, or missed something obvious. These notes are the fastest path to a sharper product.
For labs, departments, graduate programs, writing centers, and research offices that want structured pre-submission review support.
Request additional venues, specialty fields, reviewer cultures, or journal families that should be represented in the catalog.
Referee supports preparation and revision planning. It does not replace collaborators, statisticians, journal policies, or direct communication with editors.
Include the journal, field, and whether you are asking about product behavior, access, privacy, billing, or manuscript-review quality.
SUBMISSION WORKSPACES
Each manuscript has one private workspace for its repeating cycle: review the PDF, edit the Word document, check a revision, and track journal submissions—with every report and version kept together.
JOIN A LAB GROUP
Enter the 8-character invite code your PI shared with you.
SAVED REPORTS
Review history turns one-off critique into a revision trail. Compare earlier and later runs, preserve decision logic for coauthors, and keep generated reports even after deleting uploaded manuscript files.
Run a revised manuscript and compare whether the highest-risk objections changed.
Keep the report, editor-style letter, ranked fixes, and export files for lab meetings or coauthor review.
Delete manuscript uploads while retaining the generated report history tied to your account.
ACCOUNT SETTINGS
Choose a strong password of at least 8 characters.
A verification code will be sent to your new address before the change takes effect.
Enter the 6-digit code sent to your new address:
You can cancel any time. Your plan stays active until the end of the current billing period. Use the billing portal to complete cancellation.
Download a complete JSON export of your account, all generated reviews, and reported outcomes. Use this to keep a local copy or move to another tool.
Permanently removes your account, all reviews, all manuscript edit sessions, all stored manuscript text, and all usage data from our servers. Downloaded exports are not affected. This cannot be undone.
SECURITY & PRIVACY
RefereeBio is a private manuscript-preparation tool. This page explains, in plain language, how manuscript content is handled and what you should not upload.
RefereeBio does not publish your manuscript, submit it to a journal, or make it visible to other users. Reports remain in your private account history.
The uploaded review file is stored privately for up to 90 days so a linked revision can reuse the original automatically. You can delete it immediately while keeping the report, or choose "Keep past 90 days" and later restore the automatic limit. Extracted manuscript text is not separately retained in the saved review settings.
Edit sessions may keep suggestion metadata longer when you select "Keep past 90 days," but their raw manuscript text and quoted source excerpts are still wiped 180 days from when the session was created. This window is extended for as long as an active subscription is held — the clock only runs while your plan is inactive or free. You can manually delete an edit session or its manuscript content at any time from My Papers, before the 180-day window is reached.
When generating AI analysis, your manuscript text is transmitted to Anthropic's API. Anthropic's standard data retention is approximately 30 days; they do not use API submissions to train models by default. RefereeBio does not intentionally use your manuscript to train public AI models or sell manuscript content.
Only you can access your manuscripts and reports through the product. No other users can view your account. The site operator — whoever runs this server — has database-level access to all stored content as part of normal server administration. This is standard for any web service. Content is not shared with third parties except as necessary to deliver the AI analysis you request.
Deleting a report wipes all stored reviewer comment text from that review. Deleting an edit session wipes raw text and manuscript excerpts from every suggestion immediately. Deleting your account removes all reviews, reports, edit sessions, and personal data — only payment records required for financial compliance are retained. Note: server backups, Stripe's payment records, Anthropic's 30-day API copies, and OpenAlex search-request records are outside the scope of in-app deletion.
Do not upload patient-identifying information, HIPAA-covered data, FERPA records, export-controlled data, classified material, or other specially regulated content unless a separate written agreement covers it. You are responsible for having the right to upload the manuscript and for independently checking every output against your data, journal instructions, approvals, and field standards.
Read full legal, privacy & responsible-use termsLEGAL + SECURITY
Plain-language terms for a private manuscript-preparation tool. Last updated July 4, 2026.
By creating an account, uploading content, purchasing a plan, or using RefereeBio, you agree to these terms, the pricing shown at checkout, and any written agreement that applies to your account.
If you do not agree, please do not use the service or upload content. We may ask you to accept material updates to these terms.
RefereeBio provides private manuscript-preparation tools, including simulated peer review, journal-fit analysis, revision planning, response-letter drafting, and editing support.
AI output is a helpful starting point, not a fact or professional advice: please check it independently. We do not submit manuscripts, contact journals, make editorial decisions, promise publication, certify statistics, or replace professional review.
Outputs are AI-generated simulations and drafting aids. They may be incomplete, inaccurate, outdated, overly cautious, too severe, too lenient, or inconsistent with a real journal's editorial process. Users must independently review all outputs before relying on them.
Reports are not official journal reviews, publication forecasts, editor decisions, professional statistical certification, legal advice, medical advice, research-integrity clearance, or endorsement by any journal, publisher, editor, reviewer, institution, or society.
Journal, publisher, editor, society, and indexing-service names may appear so authors can reason about submission strategy, venue expectations, and manuscript fit. Those names are used descriptively only.
RefereeBio is independent and is not affiliated with, endorsed by, sponsored by, certified by, or operated by any journal, publisher, editor, reviewer, society, university, funder, or indexing service named in the product.
You keep ownership of your manuscripts and other content. You give us a limited permission to process it to deliver the service, keep your account history, support you, secure the product, and meet legal obligations.
Please upload only content you are authorized to use. Journal and publisher names are used descriptively only; RefereeBio is independent of them.
Manuscripts are treated as confidential user content. RefereeBio does not claim ownership of uploaded manuscripts and does not intentionally use private manuscripts to train public AI models.
When third-party AI providers are configured, manuscript text may be transmitted to those providers solely to generate the requested analysis or editing output, subject to the provider terms, retention rules, and data-processing settings used in the deployed environment. Do not upload content unless you are comfortable with that processing path.
RefereeBio is designed so manuscript analysis runs through the server rather than exposing AI credentials in the browser. Saved uploads are stored in the application upload directory with restricted file permissions where supported, and generated reports are tied to authenticated accounts.
Users can delete uploaded manuscript files at any time while keeping generated reports. Users can also permanently delete an entire review record (file and report together) at any time. Uploaded review manuscripts are automatically removed after 90 days unless the user explicitly extends retention; extended retention can be turned off later. Saved manuscript edit sessions are deleted after 180 days (counted from when each session was created) unless the user keeps their suggestion metadata, but raw edit-session manuscript text is wiped after 180 days regardless; the 180-day window is paused for as long as the user holds an active subscription, and users can delete an edit session or its manuscript content at any time from My Papers. Temporary parsed-source files and exports are cleaned up on a shorter schedule.
Only the authenticated account owner can access their manuscripts and reports through the product interface. The site operator has database-level access as part of server administration, which is standard for any web service. RefereeBio does not submit manuscripts to journals, does not publish uploads, does not intentionally use private manuscripts to train public models, and does not sell manuscript content. No system is risk-free, so only upload manuscripts you are authorized to process.
RefereeBio may collect account information, uploaded or pasted manuscript content, reviewer comments, generated reports, usage events, device or IP-derived security signals, payment status from the payment provider, and support messages you send.
This information is used to provide requested features, authenticate users, save history, enforce usage limits, process payments, troubleshoot errors, improve product quality, prevent abuse, comply with law, and respond to support or privacy requests. RefereeBio does not sell manuscript content.
We collect the information needed to run and protect the service: account details, content and reports, usage and security signals, payment status, and messages you send us. We use it to provide features, authenticate accounts, process payments, prevent abuse, improve the product, and comply with law.
We do not sell manuscript content or intentionally use private manuscripts to train public AI models. EU/EEA and California users may have rights to access, correct, delete, export, or limit certain processing—contact us to exercise them.
RefereeBio uses a session cookie (referee_sid) to authenticate your account after sign-in. This cookie is HttpOnly, SameSite=Lax, and expires when you sign out or the session times out. This cookie is strictly necessary for authentication and is always set when you are signed in.
Optional analytics cookies: With your consent, RefereeBio loads Google Analytics (Google LLC) to collect anonymous usage statistics such as page views and session duration. No manuscript content, account email, or identifiable personal data is sent to Google Analytics. Analytics cookies are not set unless you click "Accept analytics" in the consent banner. You can change your choice at any time by clearing localStorage in your browser settings (key: analytics_consent) and reloading the page.
The application uses browser localStorage to preserve ephemeral UI state (such as the active journal tab or review tab position) within a single session, and to store your analytics consent preference. No personal data or manuscript content is persisted in localStorage. You can clear this data at any time through your browser's storage settings without affecting your account or saved reports.
RefereeBio is intended for users who are at least 18 years old. If you are under 18, you may only use the service with the involvement and consent of a parent or legal guardian. RefereeBio does not knowingly collect personal information from children under 13. If we learn that we have collected personal information from a child under 13, we will delete it promptly. If you believe a child under 13 has provided us with personal information, please contact us at the address below.
Unless RefereeBio signs a separate written agreement specifically covering regulated data, do not upload protected health information, patient-identifiable data, HIPAA-covered data, FERPA-covered student records, classified information, export-controlled technical data, government-restricted data, or other content subject to special legal or contractual handling requirements.
RefereeBio is not a HIPAA business associate by default and the service is not intended to store or process protected health information. De-identify manuscripts and remove patient identifiers, private subject data, confidential sponsor information, and restricted data before upload.
AI calls and uploads are handled server-side, saved reports require an account, and payment details are handled by our payment provider. No online service is risk-free; please do not upload protected health information, identifiable patient data, classified, export-controlled, or similarly restricted content without a separate written agreement.
Raw review manuscript files are normally removed after 90 days unless you explicitly extend retention; you may restore the automatic limit or delete the file at any time. Generated reports and manuscript edits are retained for 180 days from when they are produced (30 days on the free tier), after which they are no longer accessible; you can export or delete them sooner at any time. You can export or delete account data in settings; limited payment, backup, and security records may remain where law or operations require.
This retention window applies while RefereeBio is operating. In the unlikely event that RefereeBio discontinues the service, users who purchased reports or credits will be given advance notice through the product or account channel, with at least 180 days from that announcement to access and export their reports before removal, subject to applicable law and circumstances outside our control.
AI calls are made from the backend, uploads are handled server-side, account access is required for saved history, and payment details are handled through the configured payment provider rather than stored directly by RefereeBio.
No web application can promise absolute security. Do not upload material you are not authorized to process, and avoid uploading data subject to unusually strict controls unless your deployment has an appropriate data-processing agreement, access-control policy, and security review.
Use RefereeBio for your own draft evaluation, revision planning, and submission preparation. You remain responsible for research accuracy, ethics, permissions, disclosures, authorship, and journal-policy compliance.
RefereeBio can identify likely reporting gaps and revision priorities, but it does not replace collaborators, statisticians, compliance teams, IRBs, ethics committees, clinical judgment, professional editorial review, journal policy review, conflict-of-interest analysis, authorship analysis, or data-integrity review.
Users remain responsible for research accuracy, ethical approvals, statistical validity, disclosure statements, authorship decisions, human-subjects protections, animal-use approvals, and journal-policy compliance.
Payments are handled through Stripe. Subscriptions renew automatically until cancelled; cancellation stops the next renewal but does not ordinarily refund a completed billing period unless checkout terms or applicable law say otherwise. Taxes, bank fees, and currency conversion may apply.
Plan credits: Credits included with monthly or annual plans are valid for one month after they are issued. Pack credits: credits from a single-manuscript pack do not expire while RefereeBio continues to operate. If we ever plan to discontinue the service or retire unused pack credits, we will give affected users substantial advance notice and a reasonable opportunity to use them, subject to applicable law and circumstances outside our control.
Free accounts may receive a limited trial allowance. Paid tiers unlock continued access after the trial limit is used. RefereeBio may update feature availability, plan names, prices, or usage limits over time, but material changes should be communicated through the product or account channel when practical.
Access can be limited, suspended, or terminated for misuse, security concerns, chargebacks, fraud risk, unlawful use, or violation of these terms.
Cancellation and refund handling depends on the configured payment provider and any written plan terms shown at checkout. Cancellation generally stops future renewal; it may not automatically refund amounts already paid unless the applicable plan terms, payment-provider rules, or law require it.
Billing questions can be sent to the contact address below. Include the account email and the plan involved, but do not send full card numbers.
The service may be modified, interrupted, rate-limited, or discontinued. Features that depend on AI providers, payment providers, file parsing, model availability, or third-party infrastructure may fail or change without notice.
RefereeBio may update this legal page as the product changes. Continued use after an update means you accept the updated terms, subject to any rights you may have under applicable law.
To the fullest extent permitted by law, the service is provided “as is” and “as available,” without a promise of accuracy, uninterrupted availability, or any publication outcome. RefereeBio and its founders, officers, employees, and agents are not responsible for indirect or consequential loss, or for decisions made from AI output.
The total liability of RefereeBio and its founders, officers, employees, and agents for any claim relating to the service is limited to the greater of what you paid us in the six months before the claim or $100. Some places limit these exclusions, so they may not apply in full to you.
The service is provided "as is" and "as available." To the fullest extent permitted by law, RefereeBio and its founders, officers, employees, and agents disclaim all warranties, express or implied, including warranties of accuracy, reliability, availability, merchantability, fitness for a particular purpose, non-infringement, uninterrupted operation, error-free output, and that any manuscript outcome will improve.
Users are responsible for reviewing, validating, and deciding whether to use any generated text, recommendation, score, checklist, cover letter, response draft, or manuscript edit.
RefereeBio's name, software, design, analysis framework, and other service materials belong to RefereeBio or its licensors. Except for your content and rights expressly granted for your own reports, you may not copy, resell, scrape, reverse engineer, or use them to offer a competing service except where law permits.
You are responsible for your content and use of the service, and agree to defend, indemnify, and hold harmless RefereeBio and its founders, officers, employees, and agents from reasonable claims arising from your unauthorized content, unlawful use, or breach of these terms.
The RefereeBio name, product design, software, analysis framework, report structure, prompts, text, graphics, trade dress, and other service materials are owned by RefereeBio or its licensors and are protected by copyright, trademark, trade secret, and other laws.
Except for rights expressly granted to users for their own reports and uploaded content, all rights are reserved. You may not copy, reverse engineer, scrape, resell, frame, or offer a competing service using RefereeBio materials except where allowed by law or written permission.
Copyright © 2026 RefereeBio. Third-party names and trademarks belong to their owners and are used descriptively.
If you believe material in the service infringes your rights, email us enough information to identify the material, your claimed rights, and how to reach you.
These terms are governed by Connecticut law. A separate signed agreement controls if it conflicts with this page, and any unenforceable provision will be limited while the rest remains effective.
Please contact us first so we can try to resolve a concern informally. Where a claim must proceed, it must be brought individually in Connecticut's state or federal courts, except where applicable law says otherwise.
Questions about privacy, access, security, billing, responsible use, or account deletion can be sent to [email protected].
Please include your account email, a brief description of the issue, and whether your question concerns privacy, billing, security, responsible use, or product behavior.
ADMIN
Use these tools when a customer contacts you about a bad review, billing issue, or needs goodwill credit.
Gift individual credits
Credits are non-expiring and stack on top of monthly limits. "All" gifts the same count for each type.
Grant temporary tier access
Grant credit bundle (review + revision check + depth edit)
Grants N reviews, N revision checks, and N depth edits. Credits are consumed per use, non-expiring.
Grant gift month (redeemable anytime)
User can redeem from their Account → Plan & Billing page at any time for 30 days of that tier.
Gift & grant history
Loading…
RESUBMISSION REVIEW
Upload your revised manuscript and the reviewer reports. Referee scores every comment, tracks what was addressed, and drafts your revision letter.
MANUSCRIPT EDITOR
Upload your manuscript as a Word document (.docx). Referee flags text-level issues and suggests careful wording changes. Science is never intentionally altered — verify all results before use.
Accept all tracked changes in your document before uploading — unresolved revisions will appear alongside Referee's suggestions in the exported file. DOCX preserves your original formatting in the export; TXT uploads export as a new plain document with tracked changes.
Once the analysis begins, feel free to close this tab and continue working. RefereeBio will carefully assess your manuscript, email you when the analysis is complete, and save the report under My Papers.
MANUSCRIPT EDIT REPORT
Review the in-text suggestions and comment blocks below.
FEEDBACK & SUGGESTIONS
Rate your experience or send an idea directly to the team. Submissions are attached to your RefereeBio account email so we can follow up when useful.
Sending as your signed-in account
This will immediately and permanently remove the following from our servers:
Payment records may be retained where required by law. Downloaded exports are not affected.
This action is permanent and cannot be undone.
To confirm, type your email address:
CONTINUE REVIEWING
THANK YOU FOR USING REFEREEBIO
Thank you for helping us test RefereeBio. Your experience and ideas are the most valuable part of this beta.
Share feedback in the feedback forum →Additional questions or comments that could improve the platform? Email [email protected].
TERMS + MANUSCRIPT SECURITY
RefereeBio is a private manuscript-preparation aid. It generates simulated review, journal-fit, revision, statistical-reporting, cover-letter, response-letter, and editing support. It does not submit manuscripts, communicate with journals, promise publication, or provide official journal guidance.
Outputs are AI-generated simulations and drafting aids. They may be incomplete, inaccurate, outdated, too severe, too lenient, or inconsistent with an actual journal process. They are not official journal reviews, publication forecasts, editorial decisions, professional statistical certification, legal advice, medical advice, research-integrity clearance, or endorsement by any journal or publisher.
You keep ownership of your manuscript and uploaded content. RefereeBio processes it to provide requested features, save account history, troubleshoot errors, secure the service, enforce usage limits, and comply with obligations. AI requests are made from the server rather than exposing AI credentials in your browser. Uploaded manuscript files can be deleted while generated reports remain available. RefereeBio does not intentionally use private manuscripts to train public AI models and does not sell manuscript content.
Manuscript text may be transmitted to configured AI, payment, hosting, email, storage, security, and analytics/service providers as needed to operate the service, subject to their terms and data-processing settings. No web service can promise absolute security.
Do not upload protected health information, patient-identifiable data, HIPAA-covered data, FERPA-covered student records, classified information, export-controlled technical data, or other specially regulated data unless RefereeBio has signed a separate written agreement covering that data. RefereeBio is not a HIPAA business associate by default.
You must have the right to upload and process the content you provide. You may not impersonate journals, editors, or reviewers; present simulated reports as official journal review; upload unauthorized content; use RefereeBio for harassment, fraud, journal manipulation, plagiarism, fabricated research, policy evasion, or misrepresentation; or rely on output without independent review.
Paid access is handled through the configured payment provider. Subscriptions may renew until cancelled. Feature availability, pricing, limits, and plans may change. Access may be limited or terminated for misuse, security concerns, fraud risk, chargebacks, unlawful use, or violation of these terms.
The service is provided "as is" and "as available" without warranties of accuracy, reliability, availability, merchantability, fitness for a particular purpose, non-infringement, uninterrupted operation, or error-free output. To the fullest extent permitted by law, the total liability of RefereeBio and its founders, officers, employees, and agents is limited to the greater of what you paid in the six months before a claim or $100, and neither RefereeBio nor its founders, officers, employees, or agents are responsible for indirect, consequential, special, punitive, lost-profit, publication-outcome, research, compliance, or editorial-decision damages. You agree to defend, indemnify, and hold harmless RefereeBio and its founders, officers, employees, and agents from claims arising from your content, misuse, legal violations, unauthorized uploads, regulated data, IP infringement, or misrepresentation of generated output.
You may request access, correction, deletion, export, or other applicable privacy choices by contacting RefereeBio. Some requests may be limited by verification, security, legal, payment, backup, or operational requirements. The full Legal, Privacy, and Responsible Use page is available from the Legal navigation tab.
ACCOUNT ACCESS
Enter your email and we'll send a link to set a new password.
ACCOUNT ACCESS
We use Google Analytics to understand how people use RefereeBio. No manuscript content is tracked. Cookie policy