Bioinformatics & statistical genomics for research groups
CBRAnalytics runs the analysis your team does not have the bench time to run: single-cell and bulk RNA-Seq, tumour-normal variant calling, and shotgun metagenomics. Every project ships with the pipelines and version-pinned containers behind the figures, so a reviewer can follow the work from FASTQ to final table.
What we run
We take on the data types that strain most core facilities, and we keep the analysis reproducible whether you send one pilot library or a multi-cohort study. Pick a track, or send us a mix.
From raw counts to the figures that go into the paper. We make the QC calls that quietly decide whether an experiment holds up: doublet removal, ambient RNA correction, batch integration, and cell-type annotation you can actually defend.
Somatic and germline calls with the artifacts worked out. We process matched tumour-normal pairs, filter against panels of normals and population databases, and annotate each call so the clinically meaningful variants separate from the sequencing noise.
Who is present, in what proportion, and whether the difference means anything. We profile 16S/ITS amplicons and shotgun libraries, then pair the composition with the diversity statistics and ordinations reviewers ask for.
For cohorts that no existing tool quite fits. We write the Nextflow or Snakemake workflow, containerise every step, and leave you with an interactive dashboard your group can keep running long after the engagement ends.
Sample deliverables
Each project ends in a set of figures and an interactive report you can share with co-authors. Below are three redacted examples, using public datasets with the sample identifiers stripped out.
14,200 peripheral blood mononuclear cells from four donors, merged and batch-corrected with Harmony. Clusters were annotated against canonical markers rather than left as numbers.
Matched tumour-normal exomes called with Mutect2 and DeepVariant, filtered against a panel of normals, then annotated so the driver events rise above the passenger noise.
Ten stool samples profiled at species and phylum level, with the community differences tested rather than eyeballed.
Every deliverable ships with the figure source, the container digest, and the exact command that produced it.
Pipeline architecture
We build each project as a versioned workflow, not a folder of scripts someone half-remembers writing. Inputs, tools, and parameters live in the workflow file, and every step runs in a pinned container.
main.nf
reproducibleRaw reads
FASTQ
QC & trimming
FastQC, fastp
Align & quantify
STAR, Salmon
Model
DESeq2, Seurat
Report
Quarto, R
$ nextflow run cbranalytics/rnaseq
[OK] 5 processes · 0 errors · report written
Data governance
Sequence data is sensitive, and we handle it that way. The commitments below are written into the contract, not tucked into an FAQ.
A mutual NDA is signed before we see a single read. It covers your samples, your metadata, and everything derived from them.
Analysis runs in an isolated environment with encryption in transit and at rest. Only the analysts assigned to your project can reach the working data.
Once you sign off, we delete the raw data on a schedule you choose, thirty days by default, and send written confirmation when it is gone.
We do not train foundation models on your data or route it through a third-party model. That is a contract term, not a setting someone can flip.
We work to data-minimisation principles and can execute a business associate agreement for clinical work. Ask during scoping and we will confirm what applies to your project.
How we work
Five steps, and you always know which one you are on.
You tell us the design: sample count, sequencing depth, and the question that needs answering. We reply within two business days with a fixed scope, a turnaround, and a price.
We sign the NDA and a short statement of work. Fixed price, milestone based. No hourly rates and no surprise invoices halfway through.
The pipeline runs with a progress note at each milestone, so you are never left guessing whether things are on track.
You receive the report, the figures, and the code, followed by a call where we walk your team through what the results do and do not support.
A window for questions while the paper is under review. If a reviewer asks why a gene was filtered out, we answer them directly.
The more you tell us about the design up front, the more precise our answer will be. If some of it is still open, say so.
Happy to sign an NDA before you share anything sensitive.
We couldn't send your request automatically.
Copy the brief and email it to us, or use the direct link below.
Prefer to book directly?
Schedule a Discovery Call