Infrastructure and software for computational biology teams

A new company usually has a computational scientist or two and no infrastructure yet. We set up the basics: somewhere to run analyses, somewhere to keep data, and a way of working that will still suit the team as it grows.

• Cloud accounts, access and budgets
• Pipeline runners and compute that scales to zero
• Raw data storage and reference data versioning
• Sample and metadata tracking
• Code repositories, CI and reproducible environments
• Sensible handling of partner and CRO data

AI assistants such as Claude can handle routine infrastructure work if they are set up for it. We write the skills and permissions that let your scientists do things themselves, such as starting a cloud instance, launching a pipeline run or checking what a project has cost, without waiting for an engineer. Limits on what each skill can touch, and what it can spend, are set in advance.

• Skills for your own infrastructure and pipelines
• Self-service compute within budgets you set
• Coding agents set up with tests and review
• Permissions that limit what an agent can change
• Helping the team build it into daily work

Most analyses start as a script or notebook written for one dataset. We turn them into pipelines that run on new batches of samples, give reproducible results, and show clearly when an output has changed.

• Nextflow and Snakemake
• Tests against known-good outputs
• Run tracking and provenance
• Seeing what each run costs
• Ingesting data from sequencing providers

Once results exist, other people need to see them. We build internal apps and dashboards so the lab and the wider company can find what they need without a scientist rerunning an analysis.

• Internal tools and dashboards
• Python and React applications
• Porting R Shiny apps to something maintainable
• Getting off low-code tools that stalled

If you already have code and infrastructure, we can make it easier to maintain: better tested, cheaper to run and simpler to deploy.

• Tests and CI for existing code
• Reducing cloud costs
• Security and dependency hygiene
• Moving self-hosted tools to the cloud

If you are setting up a computational team, or have one spending a lot of its time on infrastructure, we are happy to talk it through. If we are not the right fit, we can usually suggest someone who is.

Or email support at boxfishsoftware.com