HST Genomics · Bioinformatics analysis

Send the matrix.
Get back the figures,
the tables, and the code
that made them.

We do the analysis, not the sequencing. You get a written scope and a fixed price before any work starts — typically 5–8 working days, with six months of reviewer-driven revisions included.

Analysis only Fixed price, agreed first 5–8 working days Code included
your_data.h5ad · 2,350 cells · unlabelled

What this is

A bioinformatics team you can hire for one project.

Most groups do not need another full-time bioinformatician. They need one dataset turned into results, on a date they can plan around.

No sequencing contract

Keep your sequencing provider

We do not sell sequencing, so we never require you to buy it from us. Bring data from any platform, any provider, or a public dataset.

No minimum order

One sample is a real project

So is one slide, or one reviewer comment that needs a panel you do not have time to make.

No hourly billing

The price is agreed before we start

If the data turns out to be different from what was described, we re-quote and wait for your answer — we do not add it to the invoice afterwards.

What we analyse

Transcriptomics, in the three shapes it actually arrives in.

Single-cell RNA-seq

QC and filtering, clustering, cell-type annotation, marker genes, and differential expression done at subject level rather than per cell.

Spatial transcriptomics

Spatial domains, spatially variable genes, deconvolution against a matched reference, neighbourhood statistics, and overlays registered to the tissue image.

Bulk RNA-seq

Differential expression with the design written down first, enrichment, and figures that do not need re-drawing for the journal.

Integration across samples

Batch correction with an explicit over-correction check, because an integration that looks beautiful and erases the biology is the most common way this goes wrong.

Trajectory, abundance, communication

Pseudotime, differential abundance between conditions, and cell–cell communication — each with the assumption it rests on stated in the report.

Reviewer response

You send the comments and the data. You get the specific panels and statistics the reviewers asked for, and a note on what changes in the conclusion.

Full scope, and what we do not take on →

How it works

Four steps, and you can stop after the second.

The markers below are the files that actually move between us. Nothing starts until you have a written price in hand.

.h5ad / .mtx / .csv

You send what you have

A count matrix, a Cell Ranger or Space Ranger output folder, a Seurat or AnnData object, or a public accession. Tell us the comparison you want to make.

quote.pdf

We send scope and price

Normally within one working day: what will be done, what you will receive, the fixed price, and the delivery date. If your design cannot support the comparison, we say so here.

qc.log

We run it, then check it

Fixed checkpoints, not a vibe: thresholds derived from your data, annotation verified against markers, group comparisons aggregated to sample level before testing.

figures.pdf + .svg + .csv + .R

You get everything back

Vector figures, results tables, the code that produced them, and a methods paragraph you can paste into the manuscript.

Why it is fast, and still right

Automation does the repetitive part. It does not decide whether the answer is believable.

The pipeline, the parameter sweeps and the figure generation are automated — that is why a standard project takes days rather than months, and why the price can be fixed in advance. What stays deliberate is every point where a wrong call would survive quietly into a paper.

  • Thresholds come from your data. QC cut-offs are derived per dataset (median ± 3 MAD), not copied from a tutorial written for someone else's tissue.
  • Annotation is checked against markers. A label is only kept if the canonical markers support it; clusters that do not resolve stay numbered rather than being given a confident name.
  • Group comparisons are aggregated to sample level. Testing thousands of cells as if they were thousands of independent samples inflates significance. For n-versus-n designs we pseudobulk first.
  • Integration is stress-tested. Every batch correction is checked for whether it also removed the effect you are studying.
  • Every figure is reproducible. You get the script that made it, so a reviewer's "please re-plot without sample 4" costs minutes, not a new project.

Pricing

Published ranges. A fixed number before work starts.

Indicative ranges in USD. The written quote lands inside these unless your data needs something the range does not cover — in which case we tell you before, not after.

ServiceIndicativeUnit
Single-cell RNA-seq — standard analysis $380–650 per sample
Single-cell — advanced modules $900–3,200 per project
Spatial transcriptomics analysis $550–1,400 per slide
Bulk RNA-seq — differential expression $300–900 per project
Second-opinion re-analysis / reviewer response $450–1,800 per project

What is included, what is not, and how billing works →

Have data and a deadline?

Tell us the data type and the comparison. You get a written scope, a fixed price and a delivery date — normally within one working day.