Skip to content
Cerebra BioSystems

By team

Three teams we work with, and what we set up for each.

Most small biotechs and CROs we talk to look like one of these. If yours doesn't, the free audit is where we work out what fits.

Team type 1 of 3

Biotech with no bioinformatician

Your scientists generate the data. Nobody in-house turns it into results.

The situation

Run data sits on drives or in the sequencer's cloud account. Analysis goes to a vendor or consultant, so every experiment waits in their queue, costs a project fee and means sending your data off-site. When results come back, each follow-up question starts the cycle again.

What we set up

  • Automatic processing: when a run finishes, the right nf-core pipeline starts on its own.
  • A report for every run, with QC flags, differential expression, pathway results and standard figures.
  • A private AI workspace where scientists ask about their results and see the file and row behind every number.
  • Ongoing support from us, including a monthly allowance of analysis help from a bioinformatician.

What changes for your team

  • Routine runs are processed in your lab, without waiting in a vendor's queue.
  • Follow-up questions are answered from your own results, in plain English, with sources.
  • Sequencing data and results stay in your building.
  • Questions beyond the standard reports go to a bioinformatician on our side.
One run, before and after

Today

  1. Run finishes
  2. Data copied to a drive or uploaded
  3. Waits in a vendor's queue
  4. Results come back as files
  5. A follow-up question starts it again

With Cerebra BioSystems

Stays in your lab

  1. Run finishes
  2. Pipeline starts on its own
  3. Report and QC flags, ready to read
  4. Scientists ask the private AI, with sources

Team type 2 of 3

Biotech with one overloaded bioinformatician

One person handles every run, and routine processing eats their week.

The situation

Every run, QC check, figure and “quick question” goes through the same person. Processing fills their week, scripts live on their laptop, and the analysis only they can do keeps slipping. When they're away, everything waits.

What we set up

  • Version-pinned nf-core pipelines that start when a run lands, configured with your reference genomes and QC thresholds.
  • A run record for every run: input files and checksums, pipeline version, containers and settings.
  • A private AI workspace that answers routine QC and results questions from the pipeline outputs, with sources.
  • Hardware sized to your runs, with monitoring and tested updates handled by us.

What changes for your team

  • Routine processing and standard reports run without them.
  • Scientists check QC and look up results themselves, and every answer shows its source.
  • Results can be traced and re-run, instead of depending on one person's scripts.
  • It doesn't replace them. It takes the busywork off their plate, so their time goes to the analysis that needs their expertise.

Team type 3 of 3

CRO serving multiple sponsors

Several sponsors' data on one system, and each has to stay separate.

The situation

Each sponsor expects its data kept apart from everyone else's, a clear record of what was run on it, and deliverables in a consistent format. Doing that by hand with folders, spreadsheets and ad-hoc scripts is slow, and hard to show when a sponsor asks.

What we set up

  • A separate project per sponsor: its own results folders, permissions and AI document search.
  • Sign-in through your identity provider. The data tools check access on every request; the AI never decides who sees what.
  • An append-only audit trail of logins, pipeline runs, AI questions with their sources, and admin actions.
  • A standard report for every run, with its pipeline version and run record attached.

What changes for your team

  • Each sponsor's data stays separate, down to the AI's document search.
  • You can show what was run, with which version and settings, and who looked at the results.
  • Every run produces a report in the same format, as a starting point for sponsor deliverables.
  • Your team can use AI on sponsor data without sending it to a cloud vendor.

Problems we address

Six common problems, and what we change for each.

  • The problem

    Waiting on analysis.

    Every run queues behind one busy person, or behind nobody at all.

    How we address it

    Pipelines start when the run finishes.

    The run watcher detects a completed run, checks the sample sheet and launches the right nf-core pipeline. The report and QC flags are produced without anyone starting them.

  • The problem

    Outsourcing turnaround.

    Sending data to a vendor means a queue, a project fee and your FASTQs leaving the building.

    How we address it

    Routine analysis runs in your lab.

    Processing, differential expression, pathway results and standard figures run on your own hardware. Questions beyond the standard reports use your monthly analysis-help allowance.

  • The problem

    AI tools legal won't approve.

    Unpublished results can't go into a public chatbot, so your scientists don't use AI on them at all.

    How we address it

    AI that runs on your hardware.

    An open-weight model runs on the AI computer in your lab. Questions, data and answers stay on your network, and access follows your company login and project permissions.

  • The problem

    Growing cloud bills and per-sample fees.

    Storage costs grow with every run, and compute and analysis fees grow with every sample.

    How we address it

    Hardware you own, plus flat support.

    You buy the hardware at vendor price and pay us a flat monthly fee for support. No per-sample, per-user or per-token charges.

  • The problem

    Ad-hoc scripts nobody can re-run.

    Analysis lives in one person's scripts, with versions and settings nobody wrote down.

    How we address it

    Pinned pipelines and a record of every run.

    Community-validated nf-core pipelines run in containers with fixed versions. Each run records its inputs, checksums, versions and settings, so it can be re-run later.

  • The problem

    Results nobody can find.

    Reports sit in email threads, shared drives and old laptops.

    How we address it

    One place to look, and to ask.

    Every run's results and report are kept on your lab storage, organised by run and project. Scientists search them, along with your papers and protocols, from the private AI workspace.

What we deliver

What you get, installed and supported.

Set up once for a fixed project fee, then kept running for a flat monthly fee. Everything runs on hardware in your lab.

  • Automated nf-core pipelines

    Community-validated pipelines with pinned versions, configured for your assays, reference genomes and QC thresholds. They start on their own when a run finishes.

    • Bulk RNA-seq: available now
    • Single-cell, WGS / WES, ATAC-seq, amplicon: on request
  • QC and run reports

    A report for every run: QC flags against your thresholds, gene counts, differential expression, pathway results and standard figures.

  • Private AI workspace

    A browser-based workspace your scientists sign into with their company account. It answers from your pipeline results, and every number shows the file and row it came from.

  • Document and protocol search

    Ask questions across your papers, protocols, SOPs and ELN exports. Answers name the document they came from.

  • Hardware specification and installation

    We specify a compact AI computer and network storage sized to your runs. You buy it at vendor price; we install it, connect it to your sequencer storage and company login, and validate it against your past results.

  • Monitoring and validated updates

    Health monitoring if you enable it, security updates, and model and pipeline updates tested against known answers before you approve them.

  • Analysis-support allowance

    A monthly allowance of help from a bioinformatician, for questions beyond the standard reports.

Why your own hardware

Why it runs in your lab, not in our cloud.

Running on hardware you own is a design choice. It has real benefits, and one real trade-off.

  • Nothing leaves your network.

    Sequencing data, results, documents and AI conversations stay on hardware in your lab. In connected mode, only optional hardware-health metrics can leave.

  • Predictable cost.

    You own the hardware and pay a flat monthly fee for support, instead of meters that grow with every sample, token or seat.

  • No uploading hundreds of gigabytes.

    Runs are processed where your sequencer writes them. Nothing waits on an upload to someone else's servers.

  • Works offline, if you need it to.

    The air-gapped option runs with no internet connection at all. Updates arrive on encrypted media and are installed on-site.

  • You decide when it changes.

    Model and pipeline updates are tested against known answers first, and nothing changes until you approve it.

  • Your login, your permissions.

    Scientists sign in with the company accounts your IT team already manages, and see only the projects they belong to.

What this makes possible

What your team can do once it's running.

These are capabilities the system is built for, not promised results. How much each one changes your week depends on your runs and your team, and the free audit is where we work that out.

  • Results waiting when a run finishes.

    The pipeline, the QC check and the report start on their own, so the first look at a run doesn't depend on someone being free.

  • Scientists asking questions of their own data.

    In plain English, with the file and row behind every number. If the answer isn't in the results, the AI says so.

  • A bioinformatician with time for real analysis.

    With routine processing and quick questions handled, their time can go to study design, new methods and the hard problems.

  • Confidential data you can use with AI.

    Unpublished results, sequences and sponsor data can go into the private AI, because none of it leaves your network.

  • Results you can trace and re-run.

    Any number in a report or an answer leads back to its run, pipeline version and settings.

Find out what your lab could automate.

30 minutes, no cost. You get a written recommendation afterwards.

30 minutes · No cost · No obligation · We'll send a written recommendation afterwards.