Open-source AI for science
From protein structure to research-paper agents, AI is becoming laboratory equipment — and science demands it be open: reproducibility is the whole point. This page tracks the open-source AI projects built for research: chemistry, biology, physics, astronomy and the meta-tools that read the literature for you.
Which fields are best covered by open-source AI?
Structural biology and chemistry lead (protein models, molecular toolkits like deepchem), with research-assistant agents growing fastest across all fields.
Can I reproduce published results with these tools?
That is precisely their promise — code, weights and pipelines in the open. Check each project's health score: an actively maintained repo is far likelier to run today.
How do projects get classified as science?
Automatic keyword and topic classification over our 9,500-project graph, tuned to avoid false positives — then daily ranking by stars, momentum and health.