Before You Outsource RNA-seq Analysis: Questions, Metadata, and Deliverables
A useful RNA-seq engagement starts with an agreed question and a reviewable analysis plan. Use this practical brief to define inputs, comparisons, responsibilities and deliverables before requesting a quote.
This guide is for a project lead or technical evaluator preparing an organizational purchase of analysis services. It focuses on bulk RNA-seq. Single-cell, spatial and other designs require additional decisions and should not be assumed to fit the same brief. BioChemIntelli’s Biological Data Analysis service begins with a defined question and an assessment of the available data.
Write the decision the analysis needs to support
Start with one or two sentences stating the experimental system, intervention or contrast, and the decision that follows. For example: “We want to compare the transcriptional response of treated and vehicle-exposed cultures at one time point and prioritize findings for an independent follow-up experiment.” This defines a task without promising that a specific pathway or biomarker will emerge.
Separate the primary question from optional exploration. Adding alternative contrasts, splicing, de novo assembly or integration with proteomics can change the data requirements and workload. Ask the provider to identify which tasks are included, which require another stage, and which the current data cannot support.
Make sample identity and experimental design inspectable
| Field | Example or purpose |
|---|---|
| sample_id | A stable identifier that matches the file inventory |
| condition | Treated or vehicle; use the actual experimental labels |
| biological_unit | Donor, animal, independent culture or other unit of replication |
| batch | Extraction, preparation and sequencing batches where known |
| pairing | Which samples come from the same biological unit, if applicable |
| library information | Library method, paired/single-end reads and strandedness if known |
| reference | Organism, assembly and annotation versions, or a request to select them |
Do not replace unknown metadata with guesses. Mark an unknown library property for investigation. Distinguish biological replicates from repeated sequencing of the same library: more files do not necessarily mean more independent experimental units. If all treated samples were prepared in one batch and all controls in another, the brief must expose that confounding before a model is proposed.
The nf-core/rnaseq input documentation provides a concrete example of the connection between sample identifiers, FASTQ files and strandedness. A pipeline samplesheet is not a complete experimental-design record. The analysis team still needs the biological relationships and covariates that determine which comparisons are meaningful.
State where the work starts and what is already available
List whether you have raw reads, alignment files, gene-level counts, normalized values or only a previous report. Include file formats, approximate sizes, checksums if available, and the provenance of earlier processing. Starting from a count matrix is a different scope from processing raw reads; reusing counts also requires knowing how they were generated.
Agree on a secure transfer method after the scope and handling requirements are understood. An initial inquiry can describe sample numbers, organism and file types without attaching confidential datasets. Human-subject, proprietary or otherwise controlled data may require additional arrangements before transfer; do not infer that a routine email exchange meets those requirements.
Turn a vague request into an explicit project brief
ILLUSTRATIVE PROJECT BRIEF — not a completed BioChemIntelli engagement
Question: compare treated and vehicle cultures at one time point.
Available inputs: 8 paired-end FASTQ sample pairs and a metadata table.
Design: 4 independent cultures per condition; processing batches documented.
Primary contrast: treated versus vehicle.
Reference: agree on assembly and annotation before processing.
Review gates: input integrity, sample QC, design review, primary analysis.
Deliverables: methods, QC report, full result tables, interpretable figures,
reproducible workflow/environment record, and a discussion of limitations.
Decision: choose findings for a separately designed follow-up experiment.
Scope exclusions: no additional time points, causal claims or validation assays.The sample numbers above make the example concrete; they are not a recommendation that four replicates are sufficient for every study. Adequacy depends on the biological variation, effect sizes, design and intended inference. A feasibility review should identify limitations and, where appropriate, recommend a design discussion before further sequencing or analysis.
Agree on deliverables that expose the analysis
Request the full result tables as well as selected figures. Specify identifiers, effect-size columns, uncertainty or significance measures appropriate to the chosen method, filtering rules and adjustment for multiple testing where applicable. A list of highlighted genes alone makes it difficult to understand what was tested or excluded.
| Decision | What to agree |
|---|---|
| Quality review | Which checks are reported, who reviews concerns, and when the project pauses |
| Analysis plan | Primary contrast, relevant covariates and limits of the design |
| Reproducibility | Software versions, parameters, reference versions and executable workflow as agreed |
| Interpretation | What a figure or ranking supports, and what needs independent evidence |
| Handover | File formats, documentation, discussion meeting and agreed correction/revision scope |
Do not require a predetermined count of significant genes as a success condition. A careful analysis may identify little signal, unsuitable samples or a design limitation. Useful delivery makes those outcomes understandable and lets your team decide what to do next. This is a stronger acceptance criterion than a polished figure with no recoverable method.
Compare proposals by scope before comparing price
Early consultation is an established service practice; the Harvard Chan Bioinformatics Core describes planning support alongside analysis. That example documents the category, not a partnership or a BioChemIntelli customer result. For your project, compare proposals against the same inputs, questions, review gates and handover requirements.
BioChemIntelli reviews whether a project fits its expertise and proposes deliverables, timeline and price before work begins. The Biological Data Analysis line includes quality checks, reproducible analysis, visualizations, interpreted findings and a meeting to discuss next steps, with the workflow determined by the data and question. Begin with a short description of the project and the files you have.
References
- nf-core contributors. nf-core/rnaseq usage Official pipeline documentation Sample identifiers, read files, strandedness and reproducible workflow inputs.
- Harvard Chan Bioinformatics Core. Consulting and Analysis Official service documentation Early consultation and a scoped fee-for-service analysis are documented category practices, not BioChemIntelli customer evidence.