Budgeting bioinformatics in a grant application
What to put in the budget for analysis, how to justify it to a review panel, and the line items applicants forget until the money is already allocated.
Do this on your own data: a grant budget line needs a defensible number, not a range copied from a blog post. The cost calculator produces one for your own design, itemised: omics.hstgenomics.com/cost-calculator. A full deliverable showing what that line item buys: omics.hstgenomics.com/sample.
Want it done rather than explained? We do analysis only — send a count matrix, an h5ad or Seurat object, Cell Ranger output or a public accession and get publication-ready results with the code that produced them, at a fixed price agreed before any work starts: omics.hstgenomics.com/quote. The same numbers are available to software as JSON, no key and no sign-up: omics.hstgenomics.com/api/quote.
Sequencing costs are easy to justify because there is a quote. Analysis is where budgets are guessed, and the guess is almost always low — which is how projects end up with data nobody has time to analyse.
A defensible rule of thumb
Budget 8–15% of your sequencing spend for analysis of standard experiments, and 20–30% when the analysis is itself novel (method development, integration of several modalities, atlas-scale work). For a project spending 40,000 on single-cell sequencing, that is roughly 3,000–6,000 for standard analysis. Panels rarely challenge a number in this range; they routinely challenge a project with 40,000 of sequencing and no analysis line at all.
The line items people forget
- Revision analysis. Reviewers will ask for more. Budget one to two additional weeks of analysis per paper — this is the item most often missing.
- Storage and transfer. Raw sequencing data for a mid-sized project runs to terabytes; storage for the grant's duration plus deposition into a public repository is a real cost.
- Compute. Cloud or cluster time for alignment and large integrations. Modest, but a line item.
- A reference dataset. Spatial deconvolution needs a matched single-cell reference. If none exists publicly for your tissue, generating one is an experiment, not a formality.
- Data deposition and documentation. Preparing a submission to GEO or ArrayExpress properly takes a day of someone's time.
How to write the justification
Panels respond to specificity. Instead of "bioinformatics support", write what will be done and by whom:
"Analysis of 24 single-cell libraries (quality control, doublet removal, integration across batches, cell-type annotation, pseudobulk differential expression across conditions, and pathway enrichment), delivered with reproducible code and deposited alongside the data. Costed at commercial rates of approximately 450 per sample."
This tells a reviewer that the applicant knows what the analysis involves — which is itself evidence the project is feasible.
In-house staff or external service?
If the analysis is central to the science and recurring, budget a person. If it is standard and episodic, an external fixed-price service is usually cheaper than a fraction of a salary, and it does not evaporate when the postdoc leaves. Many applications do both: staff time for the exploratory work, a service line for the standard pipeline and for revision analyses under deadline.
One sentence that helps at review
State that analysis code will be deposited publicly and that results are reproducible from raw data. It costs nothing, it is increasingly expected by funders, and it signals rigour more cheaply than any other sentence in the methods.