Pre-Submission Manuscript Audit
Checks a manuscript for statistical power, blinding, reagent authentication and data availability before submission.
How it works
Checks a manuscript draft against common reproducibility failures before submission: whether sample sizes are justified, whether blinding and randomisation are described, whether reagents are identified precisely enough to be re-ordered, whether exact p-values and effect sizes are reported, and whether data and code availability are stated.
Published guidelines checked in your manuscript draft
Rather than ad-hoc style rules, the pre-submission audit evaluates draft Methods and Results prose against three published reporting frameworks and links each finding to its source.
- SAMPL (Lang & Altman): always active on quantitative drafts; checks exact p-values (not "p < 0.05" thresholds alone), effect sizes with 95% confidence intervals, multiple-comparison corrections (Holm, Tukey, Bonferroni, Benjamini-Hochberg/FDR), defined variability metrics (SD, SEM, CI), and named statistical software.
- ARRIVE 2.0 Essential 10 (Percie du Sert et al., PLOS Biology 2020): activated automatically when animal terms (mice, rats, murine, in vivo, IACUC) appear; checks study groups, sample size justification, inclusion/exclusion criteria, randomisation, blinding/masking, primary outcome measures, statistical tests, animal strain/sex/age, procedures, and summary results with n.
- MIQE essential subset (Bustin et al., Clinical Chemistry 2009): activated when qPCR or RT-qPCR terms appear; checks reference gene justification, standard-curve amplification efficiency, distinction between biological and technical replicates, RNA integrity/A260/280 purity, no-template and no-RT controls, and primer sequences or assay IDs.
Scope limit: automated text detection verifies whether each reporting topic is addressed in your text; it cannot judge whether a chosen statistical test or blinding protocol was scientifically valid in execution.
Frequently asked questions
What do reviewers most often flag in methods sections?
Unjustified sample sizes, missing detail on replication — particularly whether replicates were biological or technical — antibodies and cell lines identified only by name, statistical tests named without checking their assumptions, p-values given as thresholds rather than exact values, and no statement of data or code availability. Most are cheap to fix before submission and expensive to fix during review.
What is the difference between biological and technical replicates?
Biological replicates are independent samples — separate cultures, animals or preparations — that capture genuine biological variability. Technical replicates are repeated measurements of the same sample and capture only measurement error. Statistical tests treating technical replicates as independent observations inflate the apparent sample size and produce misleadingly small p-values; n should refer to biological replicates.