Pharma Workflows

Design drugs against your target's cellular network, not just its binding site.

BioGPU shows how your target's cellular network responds to every molecule your AI designs, so that biology guides the next design.

Start a target program
Illustrative · set up once, then every compound runs on it

How a program runs

One BioGPU Program Runtime for each target.

Phase 1 sets it up around your target. Phase 2 runs every compound on it.

Illustrative · cells can be yours, or we source them

Connect your AI

Two ways in: run our agent, or plug in your own.

Either way, the experiments run on the BioGPU Runtime and the biology comes back to AI as Visual Biology Tokens.

Anima's Biology Resolution Agent

Runs Anima-built workflows

  • Target
  • Pathway
  • Therapeutics
BioGPU RuntimeCellular network experiments inside cells

Your agents

Run your proprietary workflows

  • Your orchestrator agent
  • Through BioGPU Runtime Skills
What comes back

Visual Biology Tokens: the images, the biological findings, and their metadata and embeddings, ready for your models and agents.

Phase 1 · Target Biology Resolution

Resolve the cellular network around your target.

Anima's BioGPU Agent runs cellular network experiments in your cellular system until that network is resolved.

Illustrative · one loop, six steps, repeated

The network covers what your program's experiments investigated. Tool compounds reveal network responses; they do not prove direct off-target binding.

Before the first compound

The resolved network already informs your program.

Target, strategy and AI decisions can use it before any compound is tested.

Illustrative · one network, read for different decisions

  • Target decisionsCore

    • Disease relevance and causality
    • Prioritize or stop a target

  • Therapeutic strategyCore

    • Mechanism and modality
    • Resistance and combinations

  • Your own AICore

    • Ground and train models you own on experimental evidence

  • Translational researchExploratory

    • Biomarker hypotheses, pending validation

Phase 2 · Therapeutic Biology Resolution

Compound in. Cellular network effect out.

Every compound your AI designs comes back as its cellular network effect, read against the fingerprint.

Illustrative · compounds A to D
  1. 01 · DesignYour AI designs compound A.
  2. 02 · RunBioGPU runs it on your Program Runtime.
  3. 03 · RecognizeIts network effect returns as Visual Biology Tokens.
  4. 04 · RedesignYour AI designs B, then C, then D.
MechanismEfficacySelectivityCompensationSafety signalsGo / stop

The same loop for antibodies, small molecules, ADCs, RNA therapeutics, degraders and gene editing →

Biology Resolution in Action

What BioGPU has shown in disease-relevant cells.

Each case study starts from a biological question, shows the experiment BioGPU ran and the evidence it returned.

Src in healthy donor FLSSrc in RA patient FLS
Target Biology Resolution · experimental results

Rheumatoid arthritis, read in patient-derived cells.

Healthy donor vs RA patient fibroblast-like synoviocytes, compared activity by activity.

Open the case study →
MYC network corrected by experiment in the BioGPU runtime
Target Biology Resolution · blueprint to resolved network

MYC in lung cancer: a network corrected by experiment.

A two-layer blueprint, one high-content experiment in lung cancer vs normal cells, and the map it changed.

Open the case study →

Your program

Your program's data and network are yours.

Under your agreement, what your program produces belongs to you. Anima keeps the technology that produced it.

Phase 1 is a defined engagement per target; Phase 2 is recurring capacity, scaled to your throughput.

How it scales

Most of the runtime is shared. Only your target's biology is new.

Bring us your target.

Some targets already have a resolved network, for a faster start. See available targets →

Start a target program [email protected]