PerturbAI: Which Disease Genes Are Worth Pursuing?

PerturbAI is a biotechnology company combining experimental gene perturbation with AI modelling to support target discovery, prioritisation and validation. Its approach investigates how disease-linked genes influence cells and biological circuits in living organisms, generating evidence that can inform which mechanisms and potential intervention points warrant further investigation. It addresses a central bottleneck in drug discovery: moving from an association between a gene and a disease to understanding what changing that gene actually does.

The company combines scalable in vivo perturbation, altering gene activity within living organisms, with large experimental datasets called perturbation atlases. These map relationships between genes, cellular responses, biological circuits and disease. PerturbAI uses these data to train AI models intended to predict how genetic changes and drugs reshape biological systems. The distinguishing feature is the connection between experimental interventions and computational predictions, rather than reliance on observational associations alone.

For discovery workflows, this approach could provide a more structured basis for comparing candidate targets and selecting follow-up experiments. Potential benefits include clearer prioritisation of disease mechanisms and more focused experimental planning. These remain intended benefits rather than demonstrated improvements in development timelines or therapeutic outcomes: predictions and proposed targets still require appropriate experimental confirmation.

Last checked on 23 October 2026: PerturbAI announced an AI reasoning framework with OpenAI and Scripps Research on 21 October, alongside an expansion of its OpenAI collaboration into joint drug-target discovery.

What is PerturbAI?

PerturbAI is a biotechnology company combining gene-perturbation experiments in living organisms with AI modelling to investigate disease mechanisms and support target discovery and prioritisation. Its work is relevant to pharmaceutical and biotechnology research teams studying gene function and potential therapeutic interventions. Its distinguishing technical approach uses organism-scale perturbation atlases linking genes, cells, biological circuits and disease to train models intended to predict the effects of genetic changes and drugs.

Intended Use and Limitations

• PerturbAI announced a year-long research collaboration with OpenAI and Scripps Research in October 2026, developing AI agents to interpret experimental biological data and generate testable hypotheses.

• The OpenAI collaboration is expanding into joint drug-target discovery using PerturbAI’s proprietary experimental data; validated therapeutic outcomes have not been established in the announcement.

• PerturbAI reports collaborating with NVIDIA and 10x Genomics to generate and analyse its brain-wide in vivo CRISPR atlas.

• Its mouse-brain perturbation atlas and associated manuscript were publicly released, providing external researchers with access to data for scrutiny and reuse.

• The AI reasoning collaboration is supported by a joint preprint and blinded expert report comparisons, not independent clinical validation or demonstrated patient benefit.

• PerturbAI reports closing a pre-seed funding round in Q3 2025; the announcement does not disclose the amount, investors or financial runway.

• FDA clearance, CE marking and clinical-use authorisation are not specified in the reviewed official materials; the company presents its work as drug-discovery research.

• ISO certifications, HIPAA/GDPR alignment and contractual safeguards for sensitive customer data are not specified in the reviewed official materials and require direct verification before procurement.

• No acquisition, ownership change or rebrand was identified in the reviewed official materials; the OpenAI relationship is described as a collaboration.