Pre-Submission Review for Transcriptomics Papers

Transcriptomics reviewers often ask:

• are biological replicates and sample metadata strong enough?
• are batch, donor, tissue, time, processing, sequencing lane, and library effects separated?
• are raw reads and processed matrices deposited in GEO, SRA, ArrayExpress, ENA, or another appropriate repository?
• are normalization, filtering, dimensionality reduction, clustering, and differential-expression choices justified?
• are cell-type labels or states supported by markers and context?
• are pathway, regulatory, or mechanism claims validated?
• is code available enough for reanalysis?
• does the paper fit Genome Biology, NAR Genomics and Bioinformatics, RNA, Nature Communications, a disease journal, or a methods venue?

The manuscript has to make the RNA evidence reusable and interpretable.

In our pre-submission review work, transcriptomics manuscripts most often fail when the analysis looks polished but the design and data trail cannot support the biological claim.

Replication gap: samples, donors, animals, cultures, cells, or batches are not separated clearly enough.

Batch-confounding risk: disease, treatment, site, lane, tissue handling, or dissociation protocol is aligned with batch.

Cell-label overconfidence: clusters are named as definitive cell types or states without enough marker, reference, or validation support.

Pathway overreach: gene lists and enrichment terms are written as mechanism without perturbation or orthogonal validation.

Repository incompleteness: raw reads, processed matrices, metadata, code, or annotations are not ready for review.

A useful review should identify the first transcriptomics-specific objection that would make reviewers distrust the analysis.

Genome Biology states that it operates strict open access, open source, and open data policies. NAR Genomics and Bioinformatics follows data deposition, standardization, and reproducibility guidance and asks authors to complete a data availability, standardization, and reproducibility checklist at submission. Nature Portfolio repository guidance lists SRA for DNA and RNA sequencing data, GEO and ArrayExpress for gene-expression data, and dbGaP or EGA for linked genotype and phenotype data.

NCBI guidance directs authors to submit DNA or RNA-seq reads to SRA and raw plus processed files to GEO. Single-cell RNA-seq reporting guidance emphasizes metadata because reanalysis depends on sample, cell, protocol, and processing detail.

These expectations make repository readiness part of scientific readiness.

Send the manuscript, target journal, sample metadata table, experimental design, library-preparation details, sequencing metrics, preprocessing pipeline, normalization and batch plan, differential-expression outputs, marker and annotation tables, code repository plan, GEO or SRA accession plan, processed matrices, validation plan, figures, supplement, and prior reviewer comments.

For single-cell or spatial work, include dissociation or tissue-handling details, cell or spot filtering, doublet handling, clustering settings, marker support, reference mapping, batch integration, and whether raw counts are preserved for reuse.

A useful transcriptomics pre-submission review should include:

• RNA-evidence claim verdict
• design, replication, and metadata critique
• normalization, batch, and differential-expression review
• cell-type or state annotation check
• validation and pathway-interpretation review
• repository, code, and matrix readiness note
• journal-lane recommendation
• submit, revise, retarget, or diagnose deeper call

The review should not only say "deposit data." It should identify whether another analyst could reproduce the figure-level claim.

Before submission, authors often need to:

• clarify biological replication and sample grouping
• add metadata that explains donor, tissue, time, batch, and processing
• test batch confounding and sensitivity
• document normalization, filtering, and model choices
• support cell-type labels with markers and references
• validate key expression or pathway claims
• upload raw reads, processed matrices, annotations, and code
• retarget from a mechanism journal to a transcriptomics, genomics, methods, disease, or data-resource venue

These fixes make the RNA-level claim easier to trust.

A bulk RNA-seq paper needs replication, batch control, and restrained differential-expression interpretation. A single-cell paper needs cell quality control, annotation discipline, and sample-level inference rather than cell-count inflation. A spatial transcriptomics paper needs tissue handling, registration, spot or cell segmentation, and spatial validation. A time-course paper needs temporal model logic. A disease transcriptomics paper needs clinical metadata and confounder control. A multi-omics paper needs to show how RNA evidence changes the biological interpretation.

The AI manuscript review can flag whether the blocking risk is replication, batch, annotation, validation, repository readiness, or journal fit.

Use this page when the manuscript's submission risk depends on RNA expression, RNA-seq design, transcript abundance, cell states, single-cell or spatial transcriptomics, normalization, batch, differential expression, or RNA repository readiness. Use genomics review when DNA, variants, genomes, or epigenomic assays dominate. Use proteomics review when mass-spec protein evidence dominates.

That distinction keeps the page focused on the transcriptomics buyer's actual problem.

Do not submit a transcriptomics paper if batch and biology are confounded. Reviewers will quickly ask whether the expression signal is a processing artifact.

Also pause if raw and processed data are not ready. Reviewers may not rerun everything, but missing accessions and incomplete metadata signal weak reproducibility.

For single-cell papers, pause if clusters are named too confidently. A cluster label should be supported by markers, reference context, and biological plausibility.

For mechanism claims, pause if enrichment analysis is doing all the work. Differential expression can suggest a mechanism, but it usually does not prove one.