Find open-source science resources

Autonomous multi-agent AI scientist that mirrors a PhD workflow: literature review → grounded hypothesis → real ML experiments → self-critique → write-up; features a deterministic harness with frozen-metric scoring, edit allowlists, and a verified registry to make reward-hacking and hallucination impossible, plus 108 unit tests runnable without API keys or GPUs (461+ stars, MIT License, 2026)

Diffusion-based document OCR framework replacing autoregressive decoding with block-level parallel diffusion decoding, enabling high-accuracy text recognition in scientific PDFs (613+ stars, MIT License)

Cross-platform system optimizations for accelerating AlphaFold3 training with 1.73x speedup and 1.23x memory reduction

Minimalist, batteries-included repository for training video world models with diffusion-forcing, supporting long-horizon rollouts, 3D point-cloud generation, and model-predictive control with pretrained checkpoints (Simchowitz Lab, 700+ stars, MIT License, 2026)

Learning the language of protein-protein interactions

Genomic foundation model for metagenomic and genome annotation, featuring an 8k base-pair context and 500M parameters trained on 386B base pairs of eukaryotic DNA; provides expert models and a unified CLI for prokaryotic/eukaryotic coding-sequence annotation with strong performance on Genomic Benchmarks, Nucleotide Transformer tasks, and custom Gener tasks (GenerTeam, 314+ stars, MIT License)

Automatic atomic model building program for cryo-EM maps using deep learning, enabling rapid de novo protein structure determination from electron density with high accuracy (3DEM/EMBL, 169+ stars)

Python library to train, interpret, and apply deep learning models to DNA sequences, providing a unified framework for regulatory genomics with support for CNN and transformer architectures, variant effect prediction, and attribution analysis (325+ stars)

General-purpose deep learning backbone for molecular modeling

Multi-agent system with Parser-Planner-Painter architecture converting `paper.pdf` to editable `poster.pptx`, outperforms GPT-4o with 87% fewer tokens

LLM-native molecular language that represents molecules as explicit graph-based code, enabling LLMs to operate and reason on chemistry directly with 5× lower token cost and ~76-80% accuracy on novel molecules vs ~20% for SMILES; supports small molecules, polymers, and Markush structures with lossless RDKit interconversion and Claude Code/Codex agent skills (AtomFlow, arXiv:2605.16480, 281+ stars, MIT License, 2026)

End-to-end composable multi-agent framework for automating OpenFOAM-based CFD simulations from natural language prompts, managing meshing, case setup, execution, error correction, and post-processing; achieves 100% success rate on 110 FoamBench tasks with Claude Opus 4.6 through Architect-Input Writer-Runner-Reviewer agent collaboration with RAG-enhanced generation and MCP tool integration (RPI CSML, 242+ stars, MIT License)

Programmatic framework for designing state-switching proteins via backpropagation through compositional design constraints parameterized by structure prediction models; enables de novo design of allosteric regulators and fluorescent biosensors for arbitrary small-molecule analytes (79+ stars, MIT License, ICML 2026)

High-throughput PubChem client for batch queries with caching, validation, rate-limit-aware retries, and a simple CLI.

First fully open-source model achieving AlphaFold3-level accuracy with 1000x faster binding affinity prediction (MIT)

Highly scalable equivariant deep learning interatomic potentials enabling million-atom molecular dynamics simulations with ab initio accuracy, building on E(3)-equivariant architectures for large-scale atomistic modeling (mir-group, MIT License, 480+ stars)

Generative foundation model for functional antibody and nanobody design, supporting de novo generation, affinity maturation, inverse design, structure prediction, and humanization (Tencent AI4S, ICLR 2025)

End-to-end autonomous AI research engine that turns an idea into a complete LaTeX paper by dispatching real computational experiments to local GPUs or SLURM clusters, collecting actual results, generating figures/tables, and writing a data-grounded manuscript rather than LLM hallucinations (OpenRaiser, 1.5K+ stars, MIT License, 2026)

3D vision-language model for computed tomography that leverages both structured electronic health records (EHR) and unstructured radiology reports for pretraining, enabling multimodal medical understanding and radiology report generation (447+ stars, MIT License, 2026)

Automated cell type annotation tool for single-cell transcriptomics using gradient boosting and logistic regression with reference atlases, enabling standardized classification across datasets (Wellcome Sanger Institute, Nature Biotechnology 2022)

Automate downloading, opening, and parsing DrugBank.

A package for accessing data from the NIST webbook...

The Simplified Upper Level Ontology (SULO) is ontology with a minimal set of classes and relations to guide the development of a personal health knowledge graph. [from homepage]

Parallel symbolic regression network evaluating millions of expressions on GPU with automated subtree reuse, Nature Computational Science cover article (MIT, 2026)

Generates pre-miRNA and mature miRNA count tables from read alignments to pre-miRNA sequences and a gff file, both downloaded from mirBase. Produces also read coverage plots of pre-miRNAs.

The "FRamewOrk for Molecular AGgregate Excitations" enables localised QM/QM' excited state calculations in a solid state environment.

Trackastra is a transformer-based cell-tracking tool for live-cell microscopy. It links already segmented cell instances across time by predicting associations between detections. It supports greedy tracking with or without cell divisions, optional ILP-based linking, pretrained tracking models, and export of tracked masks and lineage information in Cell Tracking Challenge format.

Distributional flow matching model for robust single-cell perturbation prediction, modeling the full distribution of perturbed cellular expression profiles conditioned on control states via PAD-Transformer and multi-kernel MMD regularization; reduces MSE by 19.6% over the strongest baseline in combinatorial settings (Westlake University, 41+ stars, MIT License)

Benchmark evaluating AI agents' ability to replicate 20 ICML 2024 Spotlight/Oral papers from scratch, with 8,316 gradable tasks and author-co-developed rubrics

End-to-end molecular dynamics engine built on PyTorch, enabling differentiable simulations with neural network potentials and GPU acceleration for machine learning-accelerated molecular dynamics (MIT License, 707+ stars)

Scaling efficient, expressive, and general SE(3)-equivariant graph attention transformers for atomic systems and machine-learned interatomic potentials (MIT License, 2026)

Pretrained machine-learned force field for (bio)molecular simulations combining the fast SO3krates neural network for semi-local interactions with universal pairwise force fields for short-range repulsion, long-range electrostatics, and dispersion interactions; supports geometry optimization, NVT/NPT/NVE MD, fine-tuning, ASE calculator, and JAX-MD integration (JACS 2025, 218+ stars, MIT License)

The EMI ontology is used to structure spectrum annotation provenance by reusing the PROV-O ontology (a W3C recommendation) and sample and observation data by applying the SOSA ontology. EMI reuses the SOSA ontology as a data schema for struturing the Sample and Observation data. SOSA (Sensor, Observation, Sample, and Actuator) is a subset of SSN (Semantic Sensor Network Ontology) that is a W3C recommendation. [from homepage]

AstraZeneca's industrial-grade retrosynthetic planning tool using MCTS to recursively decompose molecules into purchasable precursors, with multi-step route scoring and support for custom one-step models (v4.0, 2024)

Improved equivariant Transformer for 3D atomic graphs (ICLR2024)

LLM agent framework for Earth Observation with 104 specialized tools across 5 functional kits

Tool to build force field input files for molecular simulation.

GFF and GTF file manipulation and interconversion.

First agentic framework for weather science, pairing an LLM with ZephyrusWorld (a code-execution environment exposing WeatherBench 2 data, geolocation, forecasting, simulation, and climatology tools) and ZephyrusBench (2,230 Q&A pairs across 49 weather-science tasks); outperforms text-only baselines by up to 44.2 percentage points (UC San Diego Rose-STL-Lab, 99+ stars, MIT License, 2026)

AlphaFold fine-tuned with flow matching for generating protein conformational ensembles, covering both experimental PDB states and molecular dynamics ensembles at physiological temperatures; includes ESMFlow variant (MIT, 526+ stars, 2024)

Transform arXiv papers into Beamer slides using LLMs

Structure-aware protein language model using 3D structural vocabulary (Foldseek) for joint sequence-structure pretraining, achieving SOTA on protein engineering and fitness prediction benchmarks (ICML 2024, Westlake University & Repl)

Scientific foundation model and AI research copilot for idea generation, cross-disciplinary connection discovery, and hypothesis formation; trained with a decoupled reward-comment RL architecture and achieves GPT-4o-competitive novelty/rationale on STEM and social-science idea-generation benchmarks (270+ stars, MIT License, 2026)

First benchmark for automatic video generation from scientific papers (NeurIPS 2025)

RFantibody is a pipeline for structure-based de novo antibody and nanobody design, integrating backbone design with RFdiffusion, sequence design with ProteinMPNN, and structure prediction with RoseTTAFold2. It provides a comprehensive toolset for generating and filtering high-quality antibody designs.

Interactive personal genome analysis toolkit using Claude Code and Python. Parses raw genotyping data from consumer DNA services and analyzes SNPs across 17 categories including health risks, pharmacogenomics, ancestry, and nutrition, with a terminal-style HTML dashboard.

Open-source platform for building, extending, and experimenting with scientific agents, providing modular agent construction tools and standardized evaluation pipelines for accelerating autonomous scientific discovery research (748+ stars, MIT License)

First benchmark evaluating LLMs' ability to rediscover scientific laws through interactive experimentation across 324 tasks in 12 physics domains, featuring memorization-resistant metaphysical shifts of canonical laws (HKUST)

Flow-matching protein folding model using only general-purpose transformer layers, scaled to 3B parameters and trained on 8.6M+ distilled structures; challenges the reliance on complex domain-specific architectures and supports PyTorch and MLX backends with model sizes from 100M to 3B parameters (985+ stars, MIT License)

FASTQ and SAM quality control using Python.