How we work
A defensible process, from first call to handover
Analytics consulting fails in predictable ways: vague scope, undocumented methods, and a deliverable nobody can re-run. Each step below exists to close one of those gaps.
Engagement flow
Six steps, no surprises
Most projects move from first call to signed scope in under two weeks.
Scoping call — 20 minutes
We ask what data you generate, where it currently goes, what reporting takes too long, and what decision is being delayed. If we are not the right fit, we say so on this call.
Written scope and fixed price
Within a few business days you receive a scope document: deliverables, method summary, assumptions, timeline, price and what is explicitly excluded. Nothing starts before you approve it.
Analysis plan before data
Once signed, we write the analysis plan — endpoints, statistical or pipeline methods, handling of missing data, figure list. You approve it. This is what stops methods from being chosen to suit results.
Secure data intake
Data is transferred by whatever channel your policy requires, logged with a provenance record, and held only for the agreed duration under NDA.
Analysis, with a checkpoint
Work proceeds against the approved plan, with a mid-project checkpoint so you see intermediate output early rather than a finished report you disagree with.
Review, revision, handover
You receive the deliverable, the scripts, the written methods and a walkthrough session. Revision rounds run inside the agreed window, then the project closes cleanly.
Quality control
Four controls on every deliverable
These are not aspirations. They are gates — work does not ship until each one is satisfied.
- Written plan first. No analysis begins without an approved plan naming the methods.
- Reproducible code always. Every figure and table regenerates from raw input by a script you keep.
- Peer review where stakes are high. Regulatory, sponsor-facing and publication work is checked by a second qualified analyst.
- Fixed pre-delivery checklist. Version, provenance, assumption checks, figure legends and file formats verified before release.
Staffing
Who does the work, and how they are chosen
Calyra BioData runs a specialist bench rather than a fixed department. You are told who is assigned to your project before it begins, and what they have done before.
Portfolio and test project
Every analyst is assessed on prior work, must demonstrate they can write an analysis plan before touching data, and completes a small paid test project before being put on a client engagement.
A bench, not one person
Two to three vetted specialists are maintained per service line, so one person's availability never becomes your missed deadline.
One point of contact
You deal with the practice, not a marketplace. Scoping, quality control and delivery responsibility stay in one place regardless of who executes.
Tools
What we build with
Standard, auditable, and portable. Nothing you would need a licence from us to keep using.
Statistics
R (tidyverse, survival, lme4), Python (statsmodels, scipy), SAS where a sponsor requires it.
Bioinformatics
Nextflow and Snakemake workflows, Bioconductor, DESeq2, limma, standard aligners and QC tooling.
Reporting
Quarto and R Markdown for reproducible reports; Power BI, Looker Studio or Streamlit for dashboards.
Handover
Version-controlled code, documented environments, and written methods in a format your QA unit can file.
Tell us where your lab data gets stuck.
A 20-minute scoping call. We map your data flow, name the bottleneck, and tell you plainly whether this is worth paying for.