Digital life world model ecosystem

Every change in life, predicted in silico

PERTURBMIND is building PerturbSeek, a digital life world model ecosystem that simulates, in silico, how life responds to every perturbation.

01Challenge

01 / The challenge

Life faces far more perturbations than experiments cover

Genes, drugs, doses, cell types, tissues and organisms combine into an almost endless space, and experiments reach only a small part of it. A model that works only on known systems falls far short; the real challenge is to learn laws that generalize to the unknown.

02 / Our approach

Our method

Drag the point and watch it flow to a new state.

  1. 01

    See cells and organisms whole

    Not a handful of highly variable genes: we start from every protein-coding gene and extend to more modalities, building a complete digital life state space.

  2. 02

    Treat perturbation as flow

    We model each perturbation as a flow field that drives how the states of living systems evolve.

  3. 03

    Learn general dynamic laws

    Our aim is to learn general dynamic laws that transfer to entirely new cellular and biological contexts, not to memorize existing experimental results.

03 / What we are building

PerturbSeek Ecosystem

Traditional AIDD centers on targets, fitting keys to known locks. We choose to simulate how the state of the whole living system changes, and to identify effective interventions under specific disease and clinical constraints.

04Applications

  • Disease mechanism analysis
  • Novel target discovery
  • New-modality drug development

PerturbSeek

02Engine

  1. Build world models
  2. Simulate state changes
  3. Identify interventions

01Foundations

  • Perturbation omics
  • Geometric deep learning
  • World model theory
Explore our technology

04 / Who we are

Across diverse disciplines, understanding life

People from these fields refine one model, as one team.

  • 01Computational biology
  • 02Machine learning
  • 03Single-cell science
  • 04Drug discovery
  • 05Product engineering

Our team’s earlier work appeared in Nature Methods, Nature Machine Intelligence, Cancer Cell and other journals; since the lab stage, we have worked closely with leading pharma and CROs in China and abroad.

Meet the team

05 / Updates

Latest progress

  1. bioRxivPreprint

    The AlphaCell preprint is on bioRxiv

  2. DeepTechReports

    DeepTech reports on AlphaCell

  3. NatureTechnology Feature

    Nature technology feature cites AlphaCell

View all

06 / Get in touch

Let’s predict life’s next change together

Whether you’re in pharma, healthcare, academia or investment, we’d love to hear from you.