Biotechnology · global
Seeking Disease Clues in Evolutionary History: Astromech Raises $20 Million to Advance Predictive Biology
Could placing different species’ longevity and disease resistance on an evolutionary timeline help identify disease risks earlier? Astromech is building models using cross-species data, but a critical validation gap remains between recovering known genes and accurately predicting the future.
Understanding how different species evolved longevity and disease resistance could give researchers clearer directions for investigating the mechanisms of human disease. Biotechnology company Astromech is attempting to tackle this question with artificial intelligence: combining cross-species genomic and evolutionary data to estimate how biological systems might change and identify regulatory clues worth pursuing experimentally.
The concept has received support from a new funding round. SiliconANGLE reported in August 2026 that Astromech had raised $20 million, led by biotech investor Bob Nelsen. Investor KBW Ventures confirmed in September that this was the same funding round, bringing the company’s total funding to $60 million at a valuation of $3.8 billion. The funds will be used to expand the research team, increase the number of species covered by its data, and build infrastructure for comparative genomic analysis.
Co-founded by Ben Lamm and George Church, Astromech combines two computational approaches: one uses deep learning to identify patterns across species, while the other works backward along the evolutionary tree to infer ancestral states, then attempts to project changes forward. According to an overview by Inverse, its goals include identifying disease risks, drug resistance, and vulnerabilities in biological systems to generate hypotheses for further research.
The model considers not only differences in DNA sequences but also when and where genes are switched on. According to technical details published by KBW Ventures, the platform integrates chromatin accessibility, gene expression, and functional annotations to reconstruct possible ancestral gene regulatory states. These ancestral states are model inferences, rather than direct measurements; the company bases its analyses on publicly available cross-species data and functional data it generates itself.
Longevity is the first demonstration topic. The company placed 46 longevity-related genes on a time-scaled evolutionary tree to track their retention and changes across different lineages. Such comparisons aim to narrow the range of candidate mechanisms, giving researchers a stronger basis for choosing experimental directions; they do not, in themselves, mean that a treatment capable of extending human lifespan has been found.
The validation disclosed so far remains primarily retrospective. KBW Ventures said the platform can recover trait-associated genes confirmed by existing research and also proposes new candidate genes. However, these reports did not provide a complete test set, error rates, or independent validation results sufficient to assess predictive accuracy. Identifying known associations and predicting unknown changes in advance are two different tests.
According to plans announced in September, Astromech will next conduct pilots with partners in health and biosecurity to test the model’s predictions prospectively. For medical research, its practical value will depend on whether these clues hold up in functional experiments and apply to human disease. Until such evidence is available, the platform’s clearer role is to help prioritize research rather than directly support clinical decisions.