upstream/ai-boost-awesome-ai-for-science/catalogue/README.md

@@ -2,9 +2,9 @@
 title: "Readme"
 task: ""
 lineage_type: import
-upstream_source: https://github.com/ai-boost/awesome-ai-for-science/blob/4d838bbe/README.md
-upstream_sha: 4d838bbe
-imported_at: 2026-09-02
+upstream_source: https://github.com/ai-boost/awesome-ai-for-science/blob/e79d81dd/README.md
+upstream_sha: e79d81dd
+imported_at: 2026-11-03
 prompt_class: catalogue
 upstream_changes: accepted
 author: upstream
@@ -83,6 +83,7 @@ validated: false
 - [Connected Papers](https://www.connectedpapers.com/) - AI-powered visual graph for exploring academic papers and discovering connected research through citation networks and semantic similarity
 - [PaSa (ByteDance)](https://github.com/bytedance/pasa) - Advanced paper search agent powered by large language models, autonomously invoking search tools, reading papers, and selecting references to deliver comprehensive and accurate results for complex scholarly queries (1.5K+ stars, Apache 2.0, 2024)
 - [paper-search-mcp](https://github.com/openags/paper-search-mcp) - MCP server, CLI, and agent skills for searching and downloading academic papers from multiple open sources (arXiv, PubMed, bioRxiv, Semantic Scholar, OpenAlex, CORE, Europe PMC, etc.) with unified, deduplicated, LLM-friendly retrieval and an OA-first download fallback chain (OpenAGS, 1.9K+ stars, MIT License, 2025)
+- [SixSentences](https://github.com/SixSentences/sixsentences) - Self-hostable research workspace keeping literature search, deduplication, screening, data analysis and manuscript writing in one project with inspectable provenance; boolean queries compile to SQL and run offline against versioned local Parquet snapshots of the corpus, so a search re-runs to the same record set (Apache-2.0, 2026)
 
 ### Data Analysis & Visualization
 - [PandasAI](https://github.com/Sinaptik-AI/pandas-ai) - Conversational data analysis using natural language
@@ -140,6 +141,7 @@ validated: false
 - [ChartCoder (ACL 2025)](https://aclanthology.org/2025.acl-long.363/) - Multimodal LLM for chart-to-code generation, 7B model outperforms larger open-source MLLMs
 - [ChartAssistant / ChartAst (ACL 2024)](https://github.com/OpenGVLab/ChartAst) - Universal chart comprehension and reasoning model
 - [Chart-to-Text Datasets](https://github.com/vis-nlp/Chart-to-text) - Large-scale chart summarization datasets for training chart description capabilities
+- [StructEval (TMLR 2025)](https://github.com/TIGER-AI-Lab/StructEval) - Benchmark for evaluating LLM-generated structural outputs (Matplotlib, Vega, SVG, TikZ, and web UI code) with rendering-based and visual consistency checks ([paper](https://openreview.net/forum?id=buDwV7LUA7), TIGER-AI-Lab)
 
 ### Scientific Visualization Tools
 - [Chat2Plot](https://github.com/nyanp/chat2plot) - Secure text-to-visualization through standardized chart specifications
@@ -177,6 +179,7 @@ validated: false
 - [PaddleOCR 3.0 (2024/2025)](https://github.com/PaddlePaddle/PaddleOCR) - Advanced OCR with PP-StructureV3 document parsing, 13% accuracy improvement, supports 80+ languages
 - [Unstructured](https://github.com/Unstructured-IO/unstructured) - Production-grade ETL for transforming complex documents into structured formats, with open-source API
 - [Marker](https://github.com/datalab-to/marker) - High-accuracy PDF→Markdown/JSON/HTML conversion, specialized for tables/formulas/code blocks with benchmark scripts
+- [Chandra OCR 2 (datalab.to, 2026)](https://github.com/datalab-to/chandra) - State-of-the-art OCR model converting images and PDFs into structured HTML/Markdown/JSON with full layout preservation, excelling at complex tables, handwritten notes, mathematical equations, forms, and 90+ languages; tops external olmOCR benchmarks and ships a pip-installable `chandra-ocr` package with HuggingFace local and vLLM server inference modes (datalab.to, the team behind Marker, 12.5K+ stars, Apache 2.0 code / OpenRAIL-M model license)
 - [S2ORC doc2json (AllenAI)](https://github.com/allenai/s2orc-doc2json) - Large-scale PDF/LaTeX/JATS parsing to standardized JSON for millions of papers
 - [GROBID](https://github.com/kermitt2/grobid) - Machine learning software for extracting structured metadata from scholarly documents
 - [Science-Parse / SPv2 (AllenAI)](https://github.com/allenai/science-parse) - Parse scientific papers to structured fields (title/author/sections/references)
@@ -193,6 +196,7 @@ validated: false
 - [scispacy (AllenAI)](https://github.com/allenai/scispacy) - Full spaCy pipeline and models for scientific/biomedical documents, enabling named entity recognition, abbreviation resolution, and UMLS linking for scientific literature mining (1.9K+ stars, Apache 2.0)
 
 ### Scientific Literature RAG & Analysis
+- [PaperFold](https://github.com/chenxiachan/paperfold) - ArXiv reader with five-level semantic zoom from overview down to verbatim excerpts, AI summaries, and paragraph-linked answers, folding papers, web pages, files, and AI replies into structured levels of detail (781+ stars, Apache 2.0, 2026)
 - [PaperQA2](https://github.com/future-house/paper-qa) - High-accuracy RAG for scientific PDFs with citation support, agentic RAG, and contradiction detection
 - [OpenScholar](https://github.com/AkariAsai/OpenScholar) - Retrieval-augmented LM synthesizing scientific literature from 45M papers with human-expert-level citation accuracy, outperforming GPT-4o by 5% on ScholarQABench (Nature 2026, UW & Ai2)
 - [Valsci](https://github.com/bricee98/Valsci) - Self-hostable scientific claim-verification and literature-review tool combining Semantic Scholar retrieval, bibliometric scoring, and LLM-based evidence synthesis for large-batch validation workflows
@@ -210,6 +214,7 @@ validated: false
 - [OpenAI4S](https://github.com/PKU-YuanGroup/OpenAI4S) - Open-source hybrid scientific research agent and workbench replicating Claude Science, combining JSON tool orchestration with persistent Python/R Code-as-Action kernels, 604 bundled science skills, MCP connectors, sandboxed local execution, and multi-provider LLM support for end-to-end scientific workflows (PKU–YuanKong Intelligence, 377+ stars, MIT License, 2026)
 - [OpenBioMed](https://github.com/PharMolix/OpenBioMed) - Open-source biomedical AI platform integrating multimodal foundation models (BioMedGPT, PharmolixFM, LangCell) with agentic workflows and 45+ Claude Code skills for drug discovery, protein engineering, and single-cell omics analysis (PharMolix & Tsinghua AIR, 1K+ stars, 2023-2026)
 - [AutoR](https://github.com/AutoX-AI-Labs/AutoR) - Human-centered research OS with terminal-first harness and local browser Studio, turning research work into reproducible artifact-backed runs through a 9-stage workflow with human approval gates, resume/rollback controls, and venue-aware manuscript packaging (1K+ stars, 2026)
+- [Dr. Claw (EMNLP 2026 System Demonstration)](https://github.com/OpenLAIR/dr-claw) - Full-stack local AI research workspace spanning literature review, ideation, experiments, academic writing, and research artifact production (OpenLAIR, 1.2K+ stars, 2026)
 - [ScholarAIO](https://github.com/ZimoLiao/scholaraio) - Agent-agnostic research infrastructure providing AI agents with a structured scientific workspace for deep PDF parsing, hybrid semantic/keyword literature search, citation-graph analysis, topic discovery, and academic writing workflows; natively integrates with Claude Code, Codex, Cursor, Cline, and AgentSkills.io (530+ stars, MIT License, 2026)
 - [BioMCP](https://github.com/genomoncology/biomcp) - Biomedical Model Context Protocol (MCP) server unifying literature search across PubMed/Europe PMC, entity pivoting across genes/variants/drugs/diseases/pathways/proteins, local study analytics, and Claude Code/Codex integration for agentic biomedical research (531+ stars, MIT License, 2025-2026)
 - [MATLAB Agentic Toolkit](https://github.com/matlab/matlab-agentic-toolkit) - Official MathWorks toolkit connecting AI agents to MATLAB via the MATLAB MCP Server and curated skills, enabling trusted engineering and scientific computing workflows with idiomatic code generation, testing, and error diagnosis in Claude Code, GitHub Copilot, OpenAI Codex, and Gemini CLI (686+ stars, BSD-3-Clause, 2026)
@@ -222,6 +227,7 @@ validated: false
 - [Wisp Science](https://github.com/xuzhougeng/wisp-science) - Open-source, local-first desktop AI research workbench for scientific computing with Python/R, MCP bioinformatics tools, SSH/WSL/GPU runtimes, and OpenAI/Anthropic models (857+ stars, 2026)
 - [Academic Research Skills (ARS)](https://github.com/Imbad0202/academic-research-skills) - Comprehensive Claude Code skill suite covering the full academic pipeline from deep research and paper writing to multi-perspective peer review, revision, and finalization; features multi-agent teams, PRISMA systematic review, style calibration, claim-level citation audits, integrity gates, and human-in-the-loop safeguards (38K+ stars, CC BY-NC 4.0, 2026)
 - [Qinyan Academic Skills](https://github.com/LeonChaoX/qinyan-academic-skills) - Curated, multilingual library of 182 installable AI agent skills for end-to-end academic research spanning literature discovery, scientific writing, grant development, bioinformatics, drug discovery, clinical research, machine learning, and data analysis (779+ stars, MIT License, 2026)
+- [Research Agent Skills](https://github.com/KalarisLabs/research-agent-skills) - 280+ open-source agent skills covering the research lifecycle — literature reviews, hypotheses, experimental design, data analysis, paper writing, citation verification, and peer review — for Claude Code, Codex, Cursor, Gemini CLI, and any Agent Skills harness (KalarisLabs, 70+ stars, 2026)
 - [SkillOpt (Microsoft, 2026)](https://github.com/microsoft/SkillOpt) - Text-space optimizer that treats agent skill documents as trainable parameters for frozen LLMs, using scored rollouts and held-out validation gates to iteratively improve reusable natural-language skills; includes SkillOpt-Sleep for nightly self-evolution and improves accuracy across Claude Code, Codex, Copilot, and direct-chat harnesses, making it a meta-tool for evolving scientific agent skill workflows (15.5K+ stars, MIT License, PyPI)
 - [Open Science (AIPOCH)](https://github.com/aipoch/open-science) - Open-source, local-first, model-agnostic AI research workbench for reproducible scientific discovery; runs Python/R notebooks, searches the web, calls scientific data connectors, and produces inspectable reports, tables, and figures in a self-hosted desktop workspace (1.5K+ stars, Apache 2.0, 2026)
 - [OmicsClaw](https://github.com/TianGzlab/OmicsClaw) - Local-first, conversational AI research partner for multi-omics analysis with CLI, desktop app, and 95+ reproducible skills; keeps raw data local while routing natural-language requests to Python/R/CLI tools with persistent memory, autonomous analysis paths, and multi-method consensus workflows (TianGzlab, 155+ stars, Apache 2.0, 2026)
@@ -240,6 +246,7 @@ validated: false
 - [SciWrite](https://github.com/labarba/sciwrite) - Agent skill for AI-assisted scientific manuscript writing review distilled from Stanford's *Writing in the Sciences* course, performing five sequential editorial audit passes on clarity, voice, structure, consistency, and integrity (2026)
 - [PaperSpine](https://github.com/WUBING2023/PaperSpine) - Motivation-driven academic writing system for Claude Code, Codex, OpenClaw, and Hermes CLI that learns from strong papers, builds evidence-aware central-argument blueprints, and rewrites manuscripts with revision matrices and LaTeX-safe audits (4.9K+ stars, MIT License, 2026)
 - [Claude Prism](https://github.com/delibae/claude-prism) - Offline-first scientific writing workspace powered by Claude, integrating LaTeX, Python, and 100+ scientific skills with local execution, Zotero integration, and privacy-focused design (2026)
+- [ScholarCopilot (COLM 2025)](https://github.com/TIGER-AI-Lab/ScholarCopilot) - Open-weight (Qwen2.5-7B) academic-writing assistant that jointly trains retrieval and generation, learning to emit a `[RET]` token that fetches and inserts accurate citations from a 500K-paper corpus with one-click BibTeX; released with open data, model weights, and a live HuggingFace demo (TIGER-AI-Lab, 253+ stars, MIT License, 2025)
 
 ---
 
@@ -261,7 +268,10 @@ validated: false
 
 ### Autonomous Research Systems (2023-2025 Breakthroughs)
 - [FunSearch (DeepMind, Nature 2023)](https://github.com/google-deepmind/funsearch) - First system to make novel, verifiable scientific discoveries by pairing LLMs with evolutionary search, solving open problems in combinatorics (cap set problem) and discovering faster matrix multiplication algorithms
+- [AlphaTensor (DeepMind, Nature 2022)](https://github.com/google-deepmind/alphatensor) - Reinforcement learning agent that discovers provably correct and faster matrix multiplication algorithms via tensor decomposition games, finding a 47-multiplication algorithm for 4x4 matrices (improving on Strassen's 49 from 1969) and releasing 14,236+ nonequivalent discovered factorizations with GPU benchmarking scripts (Google DeepMind, 2.9K+ stars, Apache 2.0)
+- [AlphaDev (DeepMind, Nature 2023)](https://github.com/google-deepmind/alphadev) - Deep reinforcement learning agent that discovers fundamental algorithms from scratch by transforming algorithm discovery into a single-player game, discovering new provably correct sorting routines up to 70% faster for short sequences and an improved hashing algorithm; the AlphaDev sorting routines were incorporated into the LLVM libc++ C++ standard library used by millions of developers (Google DeepMind, official code, 740+ stars, Apache 2.0)
 - [OpenEvolve](https://github.com/algorithmicsuperintelligence/openevolve) - Open-source implementation of AlphaEvolve's evolutionary coding agent paradigm, enabling LLMs to autonomously discover and optimize algorithms through iterative evolution, matching the approach behind DeepMind's breakthrough matrix multiplication discovery (6.2K+ stars, 2025)
+- [Darwin Gödel Machine (Sakana AI)](https://github.com/jennyzzt/dgm) - Open-ended self-improving agent that iteratively rewrites its own codebase and empirically validates each mutation on coding benchmarks (SWE-bench, Polyglot), demonstrating open-ended evolution where agents improve their ability to improve themselves, diverging into a population of diverse specialists (arXiv 2505.22954, 2.3K+ stars, Apache 2.0, 2025)
 - [SkyDiscover](https://github.com/skydiscover-ai/skydiscover) - Modular framework for AI-driven scientific and algorithmic discovery, providing a unified interface for implementing, running, and fairly comparing discovery algorithms across 200+ optimization tasks; introduces AdaEvolve and EvoX adaptive/evolutionary algorithms and natively supports OpenEvolve, GEPA, and Harbor-format benchmarks (skydiscover-ai, 568+ stars, Apache 2.0, 2026)
 - [EvoMaster (SJTU SAI, arXiv 2026)](https://github.com/sjtu-sai-agents/EvoMaster) - Foundational auto-research agent framework for agentic science at scale, providing modular agent construction, run-level self-evolution, and multiple SciMaster domain agents (ML-Master, X-Master, Browse-Master); outperforms general-purpose agents across authoritative benchmarks including the OpenAI Frontier Science Benchmark (206+ stars, Apache 2.0, 2026)
 - [Virtual Lab (Stanford Zou Group, Nature 2025)](https://github.com/zou-group/virtual-lab) - AI-human collaborative research platform where a human researcher works with a team of LLM agents via team and individual meetings to perform scientific research; demonstrated by designing new SARS-CoV-2 nanobodies with wet-lab validation
@@ -395,10 +405,12 @@ validated: false
 ### Neural Operators & Model Discovery
 - [DeepONet](https://github.com/lululxvi/deeponet) - Learning nonlinear operators
 - [PySINDy](https://github.com/dynamicslab/pysindy) - Sparse identification of nonlinear dynamics
+- [PyDMD (JOSS 2021)](https://github.com/PyDMD/PyDMD) - Python package for Dynamic Mode Decomposition (DMD) and its variants (exact, incremental, multiresolution MR-DMD, higher-order HODMD, parametric, BOP-DMD, DMD-control, and DL-DMD neural-network extensions), enabling data-driven reduced-order modeling, Koopman operator analysis, and flow-field feature extraction for high-dimensional dynamical systems in fluid dynamics and scientific computing (1.2K+ stars, MIT License, actively maintained)
 - [PySR](https://github.com/MilesCranmer/PySR) - High-performance symbolic regression for discovering interpretable scientific equations from data, multi-population evolutionary search with Python/Julia backend, widely used in physics and astronomy (Cambridge, NeurIPS 2023)
 - [LLM-SR](https://github.com/deep-symbolic-mathematics/LLM-SR) - Scientific equation discovery and symbolic regression using LLMs, combining code generation with evolutionary search (ICLR 2025 Oral)
 - [PSRN](https://github.com/x66ccff/PSRN) - Parallel symbolic regression network evaluating millions of expressions on GPU with automated subtree reuse, Nature Computational Science cover article (MIT, 2026)
 - [pykan](https://github.com/KindXiaoming/pykan) - Kolmogorov-Arnold Networks with learnable activation functions on edges instead of fixed node activations, achieving strong performance in function fitting, PDE solving, and scientific discovery with enhanced interpretability as an alternative to MLPs (MIT, 16.3K+ stars, 2024)
+- [JEPA-Anything (Gen-Verse, arXiv 2026)](https://github.com/Gen-Verse/JEPA-Anything) - Domain-agnostic predictive world-model framework built on joint-embedding predictive architectures (JEPAs), organizing latent world states into complementary factors via Orthogonal Predictive Factorization (OPF) with dedicated prediction pathways recombined for downstream readout, intervention prediction, planning, and multi-step rollout; shares one predictive core and latent state interface across cells, molecules, physical fields, control environments, and clinical trajectories, with a reusable core library, task-design tools, and HuggingFace model collection (377+ stars, Apache 2.0, 2026)
 - [Fourier Neural Operator](https://github.com/neuraloperator/neuraloperator) - Learning operators in Fourier space
 - [Poseidon](https://github.com/camlab-ethz/poseidon) - Efficient foundation models for PDEs with pretrained transformer-based neural operators and downstream task fine-tuning pipelines, HuggingFace integration for models and datasets (ETH Zurich CAMLab, arXiv 2024)
 - [GAOT (NeurIPS 2025)](https://github.com/camlab-ethz/GAOT) - Geometry Aware Operator Transformer serving as an efficient and accurate neural surrogate for PDEs on arbitrary domains, combining geometric priors with transformer architectures for scientific computing (ETH Zurich CAMLab, 92+ stars)
@@ -505,6 +517,7 @@ validated: false
 #### Protein & Drug Discovery
 - [CryoDRGN](https://github.com/ml-struct-bio/cryodrgn) - Neural network-based cryo-EM heterogeneous reconstruction, modeling continuous 3D structure distributions from single-particle images, with CryoDRGN-ET extending to in-cell cryo-electron tomography (MIT CSAIL, Nature Methods 2021/2024)
 - [ModelAngelo](https://github.com/3dem/model-angelo) - 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)
+- [Topaz (Nature Communications 2019)](https://github.com/tbepler/topaz) - Deep-learning particle-picking framework for cryo-electron microscopy using positive-unlabeled convolutional neural networks that detect protein particles in micrographs at competitive precision with far fewer labeled examples than fully supervised baselines, serving as the automated front end of cryo-EM structure determination pipelines and adopted alongside CryoSPARC/RELION workflows with pre-trained general models, micrograph denoising, and CPU/GPU inference (Bepler Lab, MIT, 222+ stars, GPL-3.0, actively maintained 2019-2026)
 - [AlphaFold](https://github.com/google-deepmind/alphafold) - Protein structure prediction
 - [AlphaFold3](https://github.com/google-deepmind/alphafold3) - AlphaFold 3 inference pipeline for unified biomolecular structure prediction of proteins, nucleic acids, small molecules, ions, and post-translational modifications (Google DeepMind, Nature 2024)
 - [AlphaFold Server](https://alphafoldserver.com/) - Free, easy-to-use web platform by Google DeepMind and Isomorphic Labs for running AlphaFold 3 predictions of biomolecular structures and interactions, enabling researchers without local infrastructure to model proteins, nucleic acids, small molecules, ions, and post-translational modifications through a searchable proteome interface (2024)
@@ -518,6 +531,7 @@ validated: false
 - [HelixFold3](https://github.com/PaddlePaddle/PaddleHelix/tree/dev/apps/helixfold3) - Baidu's open-source reproduction of AlphaFold3 in PaddlePaddle, providing pretrained weights and inference pipelines for unified biomolecular structure prediction across proteins, nucleic acids, ligands, ions, and post-translational modifications within the PaddleHelix biocomputing platform (Baidu, bioRxiv 2024)
 - [RoseTTAFold-All-Atom](https://github.com/baker-laboratory/RoseTTAFold-All-Atom) - All-atom biomolecular structure prediction for protein-nucleic acid-small molecule-metal ion complexes, enabling accurate modeling of covalent modifications and assemblies beyond proteins (Baker Lab, Science 2024)
 - [Chai-1](https://github.com/chaidiscovery/chai-lab) - Multi-modal foundation model for biomolecular structure prediction (proteins, small molecules, DNA, RNA, glycans) achieving SOTA across benchmarks, with optional MSA/template support (Chai Discovery, 2024)
+- [Chai-2 (Chai Discovery, 2025)](https://github.com/chaidiscovery/chai-lab) - Next-generation multi-modal foundation model for biomolecular structure prediction, ranking #1 at CASP16 and outperforming AlphaFold 3 on the CASP16 dataset with large gains on antibody-antigen and protein-protein complexes; open weights with free web interface (2K+ stars repo)
 - [IntelliFold](https://github.com/IntelliGen-AI/IntelliFold) - Controllable foundation model for general and specialized biomolecular structure prediction across proteins, nucleic acids, and complexes, featuring a public web server for interactive prediction workflows (IntelliGen AI, 223+ stars, Apache 2.0, 2025)
 - [SimpleFold (Apple, arXiv 2025)](https://github.com/apple/ml-simplefold) - 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)
 - [NeuralPLexer](https://github.com/zrqiao/NeuralPLexer) - State-specific protein-ligand complex structure prediction with a multi-scale deep generative model, enabling conformational state-aware modeling of molecular interactions (329+ stars, 2024)
@@ -538,6 +552,7 @@ validated: false
 - [DynamicBind (NeurIPS 2024)](https://github.com/luwei0917/DynamicBind) - Deep equivariant generative model predicting ligand-specific protein-ligand complex structures with dynamic receptor conformational flexibility, enabling accurate docking for flexible protein targets
 - [PLACER](https://github.com/baker-laboratory/PLACER) - Graph neural network operating entirely at the atomic level for protein-ligand conformational ensemble prediction and docking, generating diverse solutions through rapid stochastic denoising to model conformational heterogeneity (Baker Lab, bioRxiv 2025)
 - [targetdiff](https://github.com/guanjq/targetdiff) - 3D Equivariant Diffusion for Target-Aware Molecule Generation (ICLR2023)
+- [DiffSBDD (Nature Machine Intelligence 2024)](https://github.com/arneschneuing/DiffSBDD) - E(3)-equivariant diffusion model for structure-based drug design, jointly denoising continuous atomic coordinates and discrete atom/bond types to generate 3D molecules conditioned on the geometry of a protein binding pocket, enabling scaffold hopping and linker design workflows with strong binding-affinity and validity on CrossDocked2020 benchmarks (MIT, 534+ stars, actively cited)
 - [SeFMol (Science Advances 2026)](https://github.com/ispc-lab/SeFMol) - Semi-flexible molecular diffusion model for structure-based drug design with reinforcement learning, achieving 20× faster sampling and providing a no-code web platform for molecular design (ISPC Lab, Tongji University, 2026)
 - [ReQFlow](https://github.com/AngxiaoYue/ReQFlow) - Rectified Quaternion Flow for efficient protein backbone generation, 37× faster than RFDiffusion with 0.972 designability (ICML 2025)
 - [AlphaFlow](https://github.com/bjing2016/alphaflow) - 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)
@@ -559,6 +574,7 @@ validated: false
 - [IgGM](https://github.com/TencentAI4S/IgGM) - 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)
 - [DrugAssist](https://github.com/blazerye/DrugAssist) - LLM-based molecular optimization tool
 - [GenMol](https://github.com/NVIDIA-Digital-Bio/genmol) - ICML 2025 drug discovery generalist using masked discrete diffusion and fragment-based generation with molecular context guidance (NVIDIA)
+- [FoldBack (AISciLab, 2026)](https://github.com/AISciLab/FoldBack) - Target-aware peptide design framework that treats receptor sequence and structure as context via multimodal adapter tuning of protein language models (ESMC + ProteinMPNN features), with reinforcement-learning-based 3D dynamic feedback (ESMFold structure evaluation) to suppress unrealistic peptide conformations; ships with a systematic assessment pipeline covering peptide-target affinity, structure quality, physicochemical properties, diversity, and novelty (142+ stars, Apache 2.0, 2026)
 - [REINVENT](https://github.com/MolecularAI/Reinvent) - Industrial-grade reinforcement-learning-based generative platform for de novo molecular design with transformer architectures, supporting multi-objective optimization, scaffold decoration, and curriculum learning (AstraZeneca MolecularAI, REINVENT 4, 2024)
 - [mint](https://github.com/VarunUllanat/mint) - Learning the language of protein-protein interactions
 - [Mol-Instructions](https://github.com/zjunlp/Mol-Instructions) - Large-scale biomolecular instruction dataset for chemistry/biology LLMs (ICLR2024)
@@ -572,7 +588,9 @@ validated: false
 - [nvMolKit (NVIDIA BioNeMo, 2025)](https://github.com/NVIDIA-BioNeMo/nvMolKit) - High-performance, GPU-accelerated library for key computational chemistry tasks including molecular similarity, conformer generation, and geometry relaxation, designed to accelerate drug-discovery and molecular-modeling workflows (264+ stars, Apache 2.0)
 - [Open Targets](https://www.opentargets.org/) - Open-source data integration platform for systematic drug target identification and prioritization, combining genetics, genomics, chemistry, and pharmacology data from EMBL-EBI, Wellcome Sanger Institute, and pharmaceutical partners to accelerate therapeutic discovery
 - [ESM3](https://github.com/evolutionaryscale/esm) - 98B-parameter frontier generative model jointly reasoning over protein sequence, structure, and function, trained on 2.78 billion proteins; generated a novel fluorescent protein (esmGFP) with only 58% sequence identity to known GFPs (EvolutionaryScale, 2024)
+- [ESM Cambrian / ESMC (EvolutionaryScale & Chan Zuckerberg Biohub, arXiv 2025)](https://github.com/Biohub/esm) - Frontier protein language models (300M/600M/6B) trained on billions of protein sequences, establishing a new unsupervised scaling frontier beyond ESM2 with emergent long-range structural understanding; ships with ESMFold2 structure prediction (SOTA DockQ pass-rates on Foldbench protein-protein and antibody-antigen complexes, lab-validated de novo binder/scFv design protocol) and the ESM Atlas mapping 6.8B proteins with sparse-autoencoder-interpretable world-model features (2.9K+ stars, 2025-2026)
 - [ProtTrans](https://github.com/agemagician/ProtTrans) - State-of-the-art pretrained language models for proteins trained on thousands of GPUs and Google TPUs using Transformer architectures, enabling protein property prediction, feature extraction, and transfer learning across diverse downstream tasks (1.3K+ stars, MIT, 2020-2026)
+- [ProGen3 (Profluent, 2025)](https://github.com/profluent-ai/progen3) - Public release of Profluent's ProGen3 protein language model family, including PMC-15B supporting sequence- and structure-conditioned generation for protein design, zero-shot fitness prediction, and antibody engineering with state-of-the-art performance on fitness and docking benchmarks (114+ stars, Apache 2.0)
 - [ProstT5 (NAR Genomics and Bioinformatics 2024)](https://github.com/mheinzinger/ProstT5) - Bilingual protein language model translating between protein sequence and structure, finetuned from ProtT5-XL on 17M AlphaFoldDB structures using Foldseek's 3Di structural alphabet, enabling sequence-to-structure prediction, structure-to-sequence inverse folding, and unified protein representation learning (RostLab, 310+ stars)
 - [ESMFold](https://github.com/facebookresearch/esm) - Protein structure prediction from ESM models
 - [SaProt](https://github.com/westlake-repl/SaProt) - 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)
@@ -583,6 +601,7 @@ validated: false
 - [Foldseek](https://github.com/steineggerlab/foldseek) - Fast and accurate protein structure search using a learned 3Di structural alphabet (VQ-VAE) that discretizes tertiary interactions into structural tokens, enabling protein-universe-scale structural alignment at sequence-search speeds (4-5 orders of magnitude faster than DALI/TM-align) and underpinning many AI4S tools such as SaProt, ESMAtlas search, and AFDB clustering pipelines (Steinegger Lab, Nature Biotechnology 2023)
 - [ImmunoStruct (Nature Machine Intelligence 2025)](https://github.com/KrishnaswamyLab/ImmunoStruct) - Multimodal deep learning framework integrating peptide-MHC protein sequence, structure, and biochemical properties to predict class-I immunogenicity for infectious disease epitopes and cancer neoepitopes with cancer-wildtype contrastive learning, enabling personalized vaccine design (Krishnaswamy Lab, Yale University)
 - [mosaic](https://github.com/escalante-bio/mosaic) - Composite-objective protein design framework integrating Boltz, AlphaFold2, OpenFold3, ProteinMPNN, and ESM via JAX-based gradient optimization over continuous relaxed sequence space for multi-property binder design (319+ stars, MIT License, 2025)
+- [TxGemma (Google DeepMind, 2025)](https://huggingface.co/collections/google/txgemma-release-67dd92e931c857d15e4d1e87) - Open Gemma-2-based LLM family (2B/9B/27B) for therapeutics development, fine-tuned on 7M examples from Therapeutics Data Commons; supports classification, regression, and generation tasks across small molecules, proteins, nucleic acids, diseases, and cell lines, with chat variants for scientific dialogue and agentic integration (Agentic-Tx), matching or beating SOTA on 64/66 therapeutic tasks (open weights on HuggingFace)
 
 #### Genomics & Bioinformatics
 - [RhoFold+](https://github.com/ml4bio/RhoFold) - End-to-end RNA 3D structure prediction using RNA language model pretrained on 23.7M sequences, outperforming existing methods and human expert groups on RNA-Puzzles and CASP15 (Nature Methods 2024)
@@ -605,7 +624,9 @@ validated: false
 - [scPRINT (Nature Communications 2025)](https://github.com/cantinilab/scPRINT) - Large transformer-based single-cell foundation model pretrained on 50 million cells for robust gene network inference, expression denoising, cell embedding, and zero-shot label prediction, leveraging ESM2 protein embeddings and bidirectional transformer architecture (Cantini Lab, 148+ stars, GPL-3.0)
 - [Tahoe-x1](https://github.com/tahoebio/tahoe-x1) - Apache 2.0 single-cell foundation model family scaling to 3B parameters, pretrained on 266M cell profiles including perturbation data and released with training, embedding, and downstream benchmarking workflows for disease-relevant single-cell tasks (2025)
 - [Stack](https://github.com/ArcInstitute/stack) - Arc Institute's single-cell foundation model enabling in-context learning at inference time via a novel tabular attention architecture, trained on 150M uniformly-preprocessed cells for generalizing biological effects and generating unseen cell profiles in novel contexts (2025)
+- [CellFM (Nature Communications 2025)](https://github.com/biomed-AI/CellFM) - 800M-parameter single-cell foundation model pretrained on ~100 million human cells, the largest single-species single-cell corpus to date, using an Enhanced Retentive Network (ERetNet) with linear-time complexity instead of standard transformer attention; supports cell type annotation, perturbation response prediction, gene function prediction, and gene network analysis, with pretrained weights and LoRA fine-tuning scripts (Sun Yat-sen University, Chongqing University & Huawei, 117+ stars, CC BY-NC-ND 4.0)
 - [State (Arc Institute, bioRxiv 2025)](https://github.com/ArcInstitute/state) - Machine learning model predicting cellular perturbation response across diverse contexts with State Transition (ST) and State Embedding (SE) variants, featuring CLI tooling, PyPI distribution, and Virtual Cell Challenge integration (575+ stars)
+- [TranscriptFormer (Chan Zuckerberg Initiative, bioRxiv 2025)](https://github.com/czi-ai/transcriptformer) - Family of generative single-cell foundation models (TF-Metazoa, TF-Exemplar, TF-Sapiens) jointly modeling genes and their expression levels via expression-aware autoregressive transformers, trained on up to 112M cells across 12 species spanning 1.53 billion years of evolution; achieves robust zero-shot cell type classification across species, disease state identification in human cells, and prediction of cell-type-specific transcription factors and gene-gene regulatory relationships, pip-installable with pretrained weights (CZI, 166+ stars, MIT License)
 - [scvi-tools](https://github.com/scverse/scvi-tools) - Deep probabilistic framework for single-cell and spatial omics analysis, integrating scVI, scANVI, totalVI and other VAE-based models for batch correction, cell annotation, multi-omics integration, and RNA velocity (scverse/NumFOCUS, Nature Methods 2018/2024)
 - [CellRank](https://github.com/scverse/cellrank) - Probabilistic framework for inferring cell fate decisions and trajectory dynamics from multi-view single-cell data using Markov chains and machine learning, integrating RNA velocity, pseudotime, and metabolic labeling to predict differentiation paths and terminal states (scverse/Theis Lab, 449+ stars, BSD 3-Clause)
 - [cellxgene (Chan Zuckerberg Initiative)](https://github.com/chanzuckerberg/cellxgene) - Interactive explorer for single-cell transcriptomics data enabling visualization of UMAP/t-SNE embeddings, differential expression analysis, and cross-dataset comparison through a fast web-based interface; widely adopted for exploring atlas-scale single-cell datasets and integrating with AI/ML analysis workflows (773+ stars, MIT License)
@@ -614,6 +635,7 @@ validated: false
 - [scDFM (ICLR 2026)](https://github.com/AI4Science-WestlakeU/scDFM) - 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)
 - [scTranslator (Nature Biomedical Engineering 2025)](https://github.com/TencentAILabHealthcare/scTranslator) - Pre-trained large generative model translating single-cell transcriptomes to proteomes in an alignment-free manner, generating absent protein abundance data for CITE-seq, spatial CITE-seq, REAP-seq, and NEAT-seq across tissues and diseases; offers three model variants pretrained on 2M human cells, 160K PBMCs, or 18K bulk samples (Tencent AI Lab Healthcare, 96+ stars)
 - [scGPT](https://github.com/bowang-lab/scGPT) - Single-cell analysis with transformers
+- [UCE (Stanford SNAP, Nature 2024)](https://github.com/snap-stanford/UCE) - Universal Cell Embeddings: zero-shot single-cell foundation model pretrained on 36M cells across 11M species, learning cross-species gene function representations via a protein-language-model-informed token space; enables zero-shot cell type annotation, embedding, and integration of unseen datasets and species without fine-tuning (338+ stars, MIT License)
 - [CellWhisperer (Nature Biotechnology 2025)](https://github.com/epigen/cellwhisperer) - Multimodal AI bridging transcriptomics data and natural language, enabling intuitive chat-based exploration and analysis of single-cell RNA-seq datasets through conversational interaction without coding; fine-tuned Mistral 7B LLaVA model emulating biologist-bioinformatician discussions (207+ stars, GPL-3.0)
 - [CellTypist](https://github.com/Teichlab/celltypist) - 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)
 - [mLLMCelltype](https://github.com/cafferychen777/mLLMCelltype) - Multi-LLM consensus framework for automated cell type annotation in single-cell transcriptomics, integrating predictions from 10+ large language models with iterative discussion and uncertainty quantification to reduce single-model biases, achieving up to 95% accuracy without reference datasets; available as CRAN R package and PyPI Python package with Scanpy/Seurat integration (2025)
@@ -621,6 +643,7 @@ validated: false
 - [OmicVerse](https://github.com/Starlitnightly/omicverse) - Unified Python framework for bulk, single-cell, and spatial RNA-seq multi-omics analysis with deep learning deconvolution (VAE) and graph neural networks, bridging Bindea, Bindea, scanpy and squidpy ecosystems (Nature Communications 2024)
 - [ChatSpatial](https://github.com/cafferychen777/ChatSpatial) - MCP server enabling spatial transcriptomics analysis via natural language, integrating 60+ methods including SpaGCN, Cell2location, LIANA+, CellRank for Visium, Xenium, MERFISH platforms
 - [Enformer](https://github.com/google-deepmind/deepmind-research/tree/master/enformer) - Gene expression prediction
+- [Borzoi (Nature Genetics 2025)](https://github.com/calico/borzoi) - Deep convolutional-transformer genomic model predicting cell- and tissue-specific RNA-seq coverage directly from 524K-base DNA sequence at 32 bp resolution, unifying transcription, splicing, and polyadenylation regulation; scores non-coding variant effects across multiple regulatory layers, competitive with or outperforming specialized models (APARENT2, Pangolin) on QTL benchmarks, pretrained on 7,611 human + 2,608 mouse ENCODE/GTEx tracks with HuggingFace weights (Calico Life Sciences, 268+ stars, Apache 2.0, 2023-2026)
 - [DNABERT](https://github.com/jerryji1993/DNABERT) - DNA sequence analysis
 - [DNABERT-2 (ICLR 2024)](https://github.com/MAGICS-LAB/DNABERT_2) - Efficient foundation model and benchmark for multi-species genome understanding with context-aware nucleotide representations, improving upon DNABERT for diverse genomic task transfer learning (UIUC MAGICS Lab, 484+ stars)
 - [gReLU (Genentech, 2024)](https://github.com/Genentech/gReLU) - 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)
@@ -635,8 +658,14 @@ validated: false
 - [GENERanno (bioRxiv 2025)](https://github.com/GenerTeam/GENERanno) - 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)
 - [GENERator (bioRxiv 2026)](https://github.com/GenerTeam/GENERator) - Long-context generative genomic foundation model using 6-mer tokenization for DNA sequence modeling and generation, with v2 model families for prokaryote and eukaryote genomes and pretrained weights available on HuggingFace (GenerTeam, 460+ stars, MIT License, 2025-2026)
 - [DeepVariant](https://github.com/google/deepvariant) - Google DeepMind's deep learning analysis pipeline for calling genetic variants (SNPs and indels) from next-generation DNA sequencing data, achieving human expert-level accuracy and widely adopted in clinical genomics, population genetics, and precision medicine; pre-trained models available for multiple sequencing platforms and organismal genomes (Nature Biotechnology 2018, 3.7K+ stars)
+- [Clair3 (Nature Computational Science 2022)](https://github.com/HKU-BAL/Clair3) - Deep-learning germline small-variant caller for long-read sequencing (PacBio HiFi/ONT) combining fast pileup calling with haplotype-resolved full-alignment calling; widely adopted as a standard in the Human Pangenome Reference Consortium production pipelines and PrecisionFDA truth challenges, with pre-trained models for multiple sequencing platforms, bioconda/Docker distribution, and a companion Clair-skills agent plugin for Claude Code/Codex (HKU-BAL, 394+ stars, BSD-3-Clause, actively maintained)
 - [Casanovo](https://github.com/Noble-Lab/casanovo) - Transformer encoder-decoder for de novo peptide sequencing from tandem mass spectrometry, translating MS/MS spectra directly to peptide sequences without reference databases, enabling identification of novel peptides for immunopeptidomics, antibody repertoires, and metaproteomes (Noble Lab UW, Nature Communications 2024)
+- [InstaNovo (InstaDeep, Nature Machine Intelligence 2025)](https://github.com/instadeepai/InstaNovo) - Transformer that translates fragment ion peaks into peptide sequences for database-free de novo sequencing in large-scale proteomics, with InstaNovo+ extending it as a multinomial diffusion model that iteratively refines predicted sequences, plus InstaNovo-P for phosphoproteomics and Winnow for calibrated confidence with FDR control (130+ stars, Apache 2.0, actively maintained)
+- [DreaMS (Nature Biotechnology 2025)](https://github.com/pluskal-lab/DreaMS) - Transformer foundation model for tandem mass spectrometry (MS/MS) self-supervised on millions of unannotated spectra from the GeMS dataset via masked peak prediction and chromatographic retention-order objectives, producing 1024-dimensional molecular representations; achieves SOTA on spectral similarity, chemical property, and molecular fingerprint prediction, and powers the DreaMS Atlas annotating 201M+ MS/MS spectra for metabolomics and natural product discovery (Pluskal Lab, IOCB Prague & MIT, 211+ stars, MIT License)
+- [Sage (J. Proteome Res. 2023)](https://github.com/lazear/sage) - High-performance open-source proteomics database search engine written in Rust, converting raw mass spectra into peptide identifications via fragment-indexed searching with machine-learned PSM rescoring (LDA), retention-time prediction, chimeric/co-fragmenting spectrum search, and TMT/label-free quantification at scale; cloud-native with S3/GCS/Azure streaming, Percolator/Mokapot-compatible output, and integration into quantms and OpenMS pipelines, widely adopted in AI-driven proteomics workflows (312+ stars, MIT License, actively maintained)
 - [Dorado](https://github.com/nanoporetech/dorado) - Oxford Nanopore's official deep-learning basecaller for nanopore sequencing, converting raw electrical signals into DNA/RNA sequences with integrated modified-base (methylation) detection and efficient CPU/GPU inference; foundational tool for long-read genomics, epigenetics, and real-time sequencing analysis (nanoporetech, 846+ stars, actively maintained)
+- [geNomad (Nature Biotechnology 2023)](https://github.com/apcamargo/genomad) - Hybrid deep learning and alignment-based tool for identifying viruses, plasmids, and other mobile genetic elements in isolates, metagenomes, and metatranscriptomes, combining neural-network gene-content classifiers with nucleotide-sequence signatures; also performs viral taxonomic assignment, provirus detection in host genomes, and functional annotation, with precomputed databases of 200K+ viral and 1M+ plasmid genomes and web apps on Galaxy and NMDC EDGE (Berkeley Lab & DOE Joint Genome Institute, 334+ stars, actively maintained)
+- [VAMB (Nature Biotechnology 2021)](https://github.com/RasmussenLab/vamb) - Deep learning tool for metagenomic binning using a variational autoencoder over co-abundance and tetranucleotide k-mer composition to reconstruct non-redundant microbial genome bins (MAGs) directly from metagenomes without reference databases, generating higher-quality bins than composition- or abundance-only methods at scale; underpins large-scale microbiome catalogs including the human gut Unified Human Gastrointestinal Genome collection (UHGG) and remains actively maintained (Rasmussen Lab, Technical University of Denmark, 335+ stars, MIT License)
 
 #### Neuroscience & Behavioral Analysis
 - [DeepLabCut](https://github.com/DeepLabCut/DeepLabCut) - Markerless pose estimation of user-defined features with deep learning for all animals including humans, enabling quantitative behavioral analysis in neuroscience and ethology (Nature Neuroscience 2018, 5.6K+ stars)
@@ -649,6 +678,7 @@ validated: false
 - [TRIBE v2](https://github.com/facebookresearch/tribev2) - Meta FAIR's foundation model of vision, audition, and language for in-silico neuroscience, predicting fMRI brain responses to naturalistic multimodal stimuli (video, audio, text) through unified Transformer architecture mapped to the cortical surface (2026)
 - [Brain2Qwerty (Meta FAIR, Nature Neuroscience 2026)](https://github.com/facebookresearch/brain2qwerty) - Non-invasive decoding of typed sentences from MEG and EEG brain recordings using a convolutional encoder, transformer, and character-level language model; official code for the Nature Neuroscience paper and Meta blog post on brain-AI communication (Meta FAIR, 894+ stars, CC BY-NC 4.0, 2026)
 - [braindecode](https://github.com/braindecode/braindecode) - Deep learning software to decode EEG, ECG or MEG signals, providing standardized neural network models, preprocessing pipelines, and evaluation workflows for brain-computer interfaces and cognitive neuroscience research (1.2K+ stars, BSD 3-Clause, actively maintained)
+- [LaBraM (ICLR 2024 Spotlight)](https://github.com/935963004/LaBraM) - Large brain model for learning generic EEG representations via self-supervised pretraining on large-scale, multi-institutional EEG corpora; introduces a vector-quantized neural tokenizer that unifies heterogeneous electrode configurations into a shared neural codebook vocabulary, enabling single-model pretraining and fine-tuning across diverse montages with strong transfer to sleep staging, abnormal EEG detection, and event type classification benchmarks (674+ stars, MIT License, 2024)
 - [snntorch](https://github.com/jeshraghian/snntorch) - Deep learning with spiking neural networks in Python, providing gradient-based training of SNNs via PyTorch autodifferentiation for brain-inspired computing and neuromorphic research, with online learning capabilities and extensive tutorials (1.9K+ stars, actively maintained)
 - [nilearn](https://github.com/nilearn/nilearn) - Machine learning and statistical learning for neuroimaging in Python, providing easy-to-use tools for fMRI and MRI analysis including decoding, connectivity estimation, and parcellation with seamless scikit-learn integration (INRIA Parietal team, 1.4K+ stars)
 - [BrainIAC (Nature Neuroscience 2026)](https://github.com/AIM-KannLab/BrainIAC) - Self-supervised vision foundation model for generalized structural brain MRI analysis, pretrained on ~49,000 scans from diverse datasets and generalizing across brain age prediction, dementia/MCI classification, IDH mutation detection, glioma survival prediction, time-to-stroke estimation, MR sequence classification, and brain tumor segmentation; outperforms task-specific models especially with limited training data (Mass General Brigham & Harvard Medical School, 129+ stars)
@@ -682,6 +712,7 @@ validated: false
 - [Medical SAM3 (AIM Research Lab, arXiv 2026)](https://github.com/AIM-Research-Lab/Medical-SAM3) - Foundation model for universal prompt-driven medical image segmentation extending SAM3 to clinical imaging, supporting 2D public benchmarks and 3D training/evaluation with text and box prompts; pretrained weights available on HuggingFace (189+ stars)
 - [MedSegX](https://github.com/MedSegX/MedSegX-code) - Generalist foundation model and database for open-world medical image segmentation, enabling universal segmentation of diverse anatomical structures and pathologies with zero-shot generalization to unseen tasks and modalities (Nature Biomedical Engineering 2025)
 - [VoxTell (MIC-DKFZ, 2025)](https://github.com/MIC-DKFZ/VoxTell) - Free-text promptable universal 3D medical image segmentation foundation model enabling zero-shot segmentation of diverse anatomical structures and pathologies via natural language prompts across CT, MRI, and other volumetric imaging modalities (DKFZ, 195+ stars, Apache 2.0)
+- [LLaVA-Med (NeurIPS 2023)](https://github.com/microsoft/LLaVA-Med) - Large language-and-vision assistant for biomedicine, instruction-tuned on GPT-4-generated biomedical multimodal instruction-following data to enable conversational visual question answering over radiology, pathology, and microscopy images, establishing open recipes for adapting general vision-language models to the biomedical domain (Microsoft Research & University of Washington, 2.2K+ stars)
 - [BiomedParse](https://github.com/microsoft/BiomedParse) - Foundation model for joint segmentation, detection, and recognition of biomedical objects across nine imaging modalities, with v2 introducing BoltzFormer architecture for end-to-end 3D inference (Microsoft, Nature Methods 2025)
 - [UniBiomed (Nature Communications 2026)](https://github.com/Luffy03/UniBiomed) - Universal foundation model for grounded biomedical image interpretation, enabling comprehensive visual understanding, reasoning, and grounding across diverse biomedical imaging modalities with strong zero-shot generalization (55+ stars, Apache 2.0, 2025-2026)
 - [MIRA (NeurIPS 2025)](https://github.com/microsoft/MIRA) - Medical time series foundation model pretrained on 454B time points from heterogeneous clinical corpora spanning ICU physiological signals and hospital EHR, with continuous-time rotary positional encoding, frequency-specialized Mixture-of-Experts, and neural ODE extrapolation for zero-shot forecasting across irregular and multimodal temporal health data (Microsoft, 399+ stars, MIT License)
@@ -700,6 +731,7 @@ validated: false
 - [BiaPy](https://github.com/BiaPyX/BiaPy) - Open-source deep learning toolbox for bioimage analysis providing a unified, configuration-driven framework for 2D/3D semantic segmentation, instance segmentation, classification, denoising, super-resolution, and self-supervised learning; integrates state-of-the-art architectures including U-Net, Vision Transformers, and ConvNeXt, designed for microscopy and biomedical imaging researchers without extensive coding expertise (MIT License, actively maintained)
 - [QuPath](https://github.com/qupath/qupath) - Open-source bioimage analysis platform for digital pathology and research, featuring AI-powered cell detection, tissue classification, and whole-slide image analysis with extensible scripting and plugin architecture (1.3K+ stars, actively maintained)
 - [BioImage.IO](https://github.com/bioimage-io/core-bioimage-io-python) - Community-driven model zoo and deployment infrastructure for AI-powered bioimage analysis, enabling standardized sharing, validation, and cross-platform execution of deep learning models across Fiji, Ilastik, napari, and other scientific imaging tools (EPFL, EMBL, and global collaborators, actively maintained)
+- [CAREamics (Jug Lab, EMBL / Human Technopole)](https://github.com/CAREamics/careamics) - Deep-learning library for AI microscopy image restoration implementing CARE (content-aware restoration, Nature Methods 2018), Noise2Void/Noise2Fast self-supervised denoising, and related pixel-restoration algorithms as a unified, pip-installable training/inference API on a PyTorch Lightning backend, with Fiji plugin and napari integration for denoising, deconvolution, super-resolution, and virtual-staining workflows across light and electron microscopy (successor to CSBDeep, 157+ stars, BSD-3-Clause, actively maintained, 2023-2026)
 
 ### ⚛ Chemistry & Materials
 
@@ -711,21 +743,27 @@ validated: false
 #### Materials Discovery
 - [GNoME](https://github.com/google-deepmind/materials_discovery) - DeepMind's graph neural network for materials exploration, discovering 2.2M new crystal structures (380K most stable) equivalent to 800 years of traditional research, with 520K+ materials dataset open-sourced (Nature 2023)
 - [FAIRChem (OMat24)](https://github.com/FAIR-Chem/fairchem) - Meta's comprehensive ML ecosystem for materials/chemistry with 118M+ DFT calculations, EquiformerV2 models achieving top Matbench Discovery performance
+- [UMA (Meta FAIR, 2025-2026)](https://huggingface.co/facebook/UMA) - Universal Model for Atoms: single pretrained universal interatomic potential jointly trained on 235M+ DFT calculations spanning inorganic materials (OMat24, sAlex) and molecules (OMol25), delivering quantum-accurate energies and forces across materials, molecules, catalysts, polymers, and biomolecules; UMA-1.2 is ~50% faster and ~40% more accurate on molecules, with open weights, pip-installable `fairchem-core`, and LAMMPS/multi-GPU inference (Meta FAIR, 2026)
 - [Skala 1.1 (Microsoft Research, 2026)](https://github.com/microsoft/skala) - Neural network-based exchange-correlation functional for density functional theory (DFT) that surpasses state-of-the-art hybrid functionals in accuracy for main-group thermochemistry, kinetics, and non-covalent interactions at semi-local DFT cost; includes PySCF/GPU4PySCF/ASE bindings and C++/Fortran integrations (248+ stars, MIT License)
 - [All-atom Diffusion Transformers (ADiT)](https://github.com/facebookresearch/all-atom-diffusion-transformer) - Unified latent diffusion transformer that jointly generates periodic crystals and non-periodic molecules, scaling to 500M parameters with SOTA results on QM9, MP20, and GEOM-DRUGS (Meta FAIR, ICML 2025, 310+ stars)
 - [JARVIS](https://github.com/usnistgov/jarvis) - NIST's open-source platform for data-driven atomistic materials design, integrating DFT datasets (JARVIS-DFT), machine learning property prediction (JARVIS-ML), and a comprehensive leaderboard for benchmarking materials AI methods across the periodic table (384+ stars)
 - [NVIDIA ALCHEMI Toolkit](https://github.com/NVIDIA/nvalchemi-toolkit) - Developer toolkit for accelerating training and inference for AI in chemistry and material science, providing optimized GPU-accelerated workflows for molecular and materials machine learning (NVIDIA, 2026)
 - [CatGo (UCSD)](https://github.com/Hello-QM/catgo-LRG) - AI-driven desktop workbench for computational materials science with an interactive 3D structure editor, natural-language CatBot assistant, visual DAG workflow engine, remote-cluster access, and HPC job submission for VASP, ORCA, CP2K, Quantum ESPRESSO, GPAW, DFTB+, SIESTA, and LAMMPS (172+ stars, AGPL-3.0, 2026)
 - [NequIP](https://github.com/mir-group/nequip) - E(3)-equivariant neural network interatomic potentials achieving DFT accuracy with up to 1000× less training data than invariant models, foundational architecture behind MACE and Allegro (Harvard, MIT, Nature Communications 2022)
+- [Nequix (Meta FAIR, NeurIPS 2025 AI4Mat / ICML 2026)](https://github.com/atomicarchitects/nequix) - Equivariant foundation models for materials "on a budget", releasing pre-trained universal interatomic potentials (nequix-mp-1 on MPtrj, nequix-omat-1 on OMat24, nequix-oam-1 combining OMat24, sAlex, and MPtrj frontier datasets) with phonon fine-tuning (PFT) for accurate lattice dynamics and analytical Hessians; pip-installable with ASE calculator and JAX/PyTorch backends with OpenEquivariance kernels (Atomic Architects, 76+ stars, MIT License, 2025-2026)
 - [Allegro](https://github.com/mir-group/allegro) - 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)
 - [SchNetPack](https://github.com/atomistic-machine-learning/schnetpack) - PyTorch toolkit for deep neural networks in atomistic simulations, implementing SchNet, DimeNet++, PaiNN, and GemNet for molecular dynamics and quantum chemistry (900+ stars)
+- [TorchANI (J. Chem. Inf. Model. 2020)](https://github.com/aiqm/torchani) - Official free and open-source PyTorch implementation of the ANI family of neural network interatomic potentials, shipping pre-trained ANI-1ccx/ANI-1x/ANI-2x ensembles with energy and force training, built-in unit conversion, and ASE integration; the foundational ANI ecosystem widely used for organic and biomolecular ML force field workflows (AIQMS, 560+ stars, MIT License, actively maintained)
 - [pymatgen](https://github.com/materialsproject/pymatgen) - Python Materials Genomics: robust materials analysis library defining classes for structures and molecules with support for many electronic structure codes; foundational toolkit powering the Materials Project (Berkeley Lab, 1.8K+ stars)
 - [MACE](https://github.com/ACEsuit/mace) - Machine learning interatomic potentials
+- [DeepMD-kit (Science Advances 2018 / JCP 2022)](https://github.com/deepmodeling/deepmd-kit) - Deep learning package for many-body potential energy representation and molecular dynamics, implementing the Deep Potential framework with GPU-accelerated model interfaces to LAMMPS, GROMACS, AMBER, OpenMM, and i-PI; ships the DPA-2 pretrained universal interatomic potential (NeurIPS 2024) fine-tunable across materials, molecules, and biomolecules, and anchors the open DeepModeling ecosystem (DeepModeling, 2.1K+ stars, LGPL-3.0, actively maintained)
+- [matgl (Materials Virtual Lab)](https://github.com/materialyzeai/matgl) - Graph deep learning library for materials science powering the M3GNet universal interatomic potential across periodic-table elements, with property prediction, structure relaxation, and crystal generation workflows built on PyTorch and DGL (576+ stars, BSD-3-Clause, actively maintained)
 - [CHGNet](https://github.com/CederGroupHub/chgnet) - Universal pretrained neural network potential with charge and magnetic moment awareness, trained on 1.5M+ Materials Project inorganic structures for charge-informed molecular dynamics and phase diagram prediction (Berkeley, Nature Machine Intelligence 2023 Cover)
 - [MatterGen](https://github.com/microsoft/mattergen) - Diffusion-based generative model for inorganic materials design, steering generation by chemistry, symmetry, bulk modulus, band gap, or magnetic properties, 2× more likely to produce stable novel structures than prior methods, experimentally validated with synthesized TaCr₂O₆ (Microsoft, Nature 2025)
 - [MatterSim](https://github.com/microsoft/mattersim) - Deep learning atomistic model across elements, temperatures, and pressures
 - [ORB](https://github.com/orbital-materials/orb-models) - Universal machine learning interatomic potential for atomistic simulation of materials, molecules, and biomolecules across the periodic table, with open-source pretrained models and inference tools (Orbital Materials, 2024-2025)
 - [SevenNet (JCTC 2024)](https://github.com/MDIL-SNU/SevenNet) - Graph neural network interatomic potential package supporting efficient multi-GPU parallel molecular dynamics simulations, enabling large-scale atomistic modeling with machine learning potentials (MDIL-SNU, MIT License)
+- [TorchMD-NET (ICLR 2022)](https://github.com/torchmd/torchmd-net) - PyTorch framework for training neural network interatomic potentials with the Equivariant Transformer (ET) architecture and its efficient TensorNet successor, providing equivariant message passing with linear complexity in tensor order; underpins the MACE-OFF and SPICE models and widely adopted across molecular dynamics and materials simulation workflows (Amsterdam Machine Learning Lab / De Fabritiis Group, 483+ stars, MIT License, actively maintained)
 - [LLaMat (Nature Machine Intelligence 2026)](https://github.com/M3RG-IITD/llamat) - Family of large language models for materials research via continued pretraining of LLaMA-2/3 on ~30B materials science tokens, outperforming commercial LLMs on materials science tasks while identifying "adaptation rigidity" in overtrained models; includes MatNLP benchmark and CIF crystal generation capabilities (IIT Delhi M3RG, MIT License)
 - [Crystal Graph CNNs](https://github.com/txie-93/cgcnn) - Crystal property prediction
 - [MatBench](https://github.com/materialsproject/matbench) - Materials informatics benchmark
@@ -739,6 +777,7 @@ validated: false
 #### Lab Automation & Robotics
 - [PyLabRobot](https://github.com/PyLabRobot/pylabrobot) - Interactive and hardware-agnostic SDK for laboratory automation, enabling programmatic control of liquid handlers, plate readers, and other lab instruments across multiple vendors; foundational infrastructure for self-driving laboratories and AI-driven experimental execution (447+ stars)
 - [RoboChem-Flex](https://github.com/Noel-Research-Group/Robochem_Flex) - Low-cost, modular self-driving laboratory platform democratizing autonomous chemical experimentation with open control software, device CAD/PCB files, and example optimization campaigns (Noël Research Group, University of Amsterdam, Apache 2.0, 2026)
+- [BayBE (EMD Group / Merck KGaA)](https://github.com/emdgroup/baybe) - Open-source Bayesian optimization and design-of-experiments framework serving as the optimization back end of self-driving laboratory campaigns, including the AlphaFlow autonomous synthesis platform (Nature 2024); provides surrogate models, active/transfer learning strategies, chemistry-aware encodings (RDKit fingerprints, descriptors), and botorch-based uncertainty handling with a unified, pip-installable API (513+ stars, Apache 2.0, 2023-2026)
 
 ### 🌌 Physics & Astronomy
 
@@ -746,6 +785,9 @@ validated: false
 - [AlphaQubit](https://github.com/google-deepmind/alphaqubit) - Google DeepMind and Google Quantum AI's transformer-based neural-network decoder for quantum error correction, trained on real Sycamore quantum processor data to outperform tensor-network and correlated matching decoders at code distances 3 and 5, demonstrating ML's role in enabling fault-tolerant quantum computing (Nature 2024)
 - [FermiNet](https://github.com/google-deepmind/ferminet) - DeepMind's neural network for ab-initio quantum chemistry, directly solving the many-electron Schrödinger equation via variational Monte Carlo with antisymmetric wavefunctions, extended to excited states (Phys. Rev. Research 2020, Science 2024)
 - [NetKet](https://github.com/netket/netket) - Machine learning toolkit for many-body quantum systems, implementing neural quantum states, variational Monte Carlo, and tensor network algorithms to solve ground-state and dynamical problems in condensed matter physics and quantum chemistry (EPFL & collaborators, Nature Physics 2019/2022+, 670+ stars)
+- [DeepQMC](https://github.com/deepqmc/deepqmc) - Open-source JAX-based software suite for variational optimization of deep-learning molecular wave functions, solving electronic ground and excited states via neural-network trial wave functions with configurable FermiNet, PauliNet, Psiformer, LapNet, and DeepErwin ansätze, geometric transferability across molecular configurations, and effective core potential support (FU Berlin / Noé group, J. Chem. Phys. 2023, 420+ stars, MIT License)
+- [ParT / Particle Transformer (Nature Machine Intelligence 2024)](https://github.com/jet-universe/particle_transformer) - Transformer architecture for high-energy physics that treats particle jets as unordered sets of constituent particles, using pairwise particle-particle attention to capture interparticle interactions; achieves state-of-the-art jet tagging on the large-scale JetClass benchmark and is widely adopted in LHC flavor-tagging pipelines (Meta FAIR & Penn State, 135+ stars, MIT License, 2023-2024)
+- [GraphNeT (JOSS 2022)](https://github.com/graphnet-team/graphnet) - Open-source Python framework for deep-learning event reconstruction at neutrino telescopes, converting sparse Cherenkov-detector readouts into graph-structured inputs for graph neural networks; supports arbitrary detector geometries (IceCube, KM3NeT, ESSνSB, Baikal-GVD) with inference orders of magnitude faster than traditional reconstruction, and powers published IceCube analyses including low-energy neutrino classification and extended northern-sky point-source searches (IceCube Collaboration & partners, 118+ stars, Apache 2.0, actively maintained)
 - [JAX-MD](https://github.com/jax-md/jax-md) - Molecular dynamics in JAX
 - [Neural ODEs](https://github.com/rtqichen/torchdiffeq) - Differential equations with neural networks
 - [Physics-Informed Neural Networks](https://github.com/maziarraissi/PINNs) - Physics-constrained ML
@@ -760,6 +802,7 @@ validated: false
 #### Astronomy & Astrophysics
 - [AstroCLIP](https://github.com/PolymathicAI/AstroCLIP) - Cross-modal self-supervised foundation model for galaxies by Polymathic AI, jointly embedding multi-band galaxy imaging and optical spectra into a shared latent space to enable zero/few-shot redshift estimation, galaxy property prediction, morphology classification, and cross-modal similarity search (MNRAS Letters 2024)
 - [AION (arXiv 2025)](https://github.com/PolymathicAI/AION) - Polymathic AI's large omnimodal foundation model for astronomical surveys, seamlessly integrating 39 distinct data modalities including imaging, spectra, photometry, and catalog entries for similarity search, property prediction, and generative modeling across legacy surveys (MIT)
+- [ML4GW (A3D3 Institute, JOSS 2025)](https://github.com/ML4GW/ml4gw) - PyTorch library for training neural networks on gravitational-wave physics, providing differentiable PSD estimation, whitening, SNR calculation, interferometer response projection, waveform simulation, and streaming data loaders; the shared back end of the NSF A3D3/ML4GW pipelines deployed for real-time detection of compact-binary coalescences in LIGO–Virgo–KAGRA observing runs (35+ stars, GPL-3.0, 2022-2026)
 - [AstroPy](https://github.com/astropy/astropy) - Python astronomy tools
 - [Gaia Archive](https://gea.esac.esa.int/archive/) - Stellar data for ML
 - [DeepSphere](https://github.com/deepsphere/deepsphere-pytorch) - Spherical CNNs for astronomy
@@ -768,14 +811,20 @@ validated: false
 
 #### Climate Modeling
 - [GenCast](https://github.com/google-deepmind/graphcast) - Google DeepMind's diffusion-based ensemble weather forecasting model at 0.25° resolution, outperforming ECMWF ENS on 97.2% of targets up to 15 days ahead, with open-source code and weights (Nature 2024)
+- [WeatherNext (Google DeepMind, 2025-2026)](https://github.com/google-deepmind/weathernext) - Family of operational-quality open weather models from DeepMind and Google Research, including WeatherNext Graph (deterministic GNN medium-range forecasting, published as GraphCast), WeatherNext Gen (diffusion ensemble, published as GenCast), WeatherNext 2 (state-of-the-art global medium-range and cyclone forecasting skillful beyond 15 days, operational at 0.25° resolution), and WeatherNext Cyclones (breakthrough tropical cyclone track forecasting, Nature 2026); official code and open weights (Apache 2.0, 7.6K+ stars)
 - [Aurora](https://github.com/microsoft/aurora) - Microsoft's foundation model for the Earth system supporting weather, air pollution, and ocean wave forecasting at multiple resolutions, trained on 1M+ hours of diverse atmospheric data (Nature 2025)
+- [Aardvark Weather (Cambridge MLIP, Nature 2025)](https://github.com/anna-allen/aardvark-weather-public) - First end-to-end data-driven weather prediction system learning directly from raw, heterogeneous Earth observations rather than physics-based reanalysis, producing both global gridded and arbitrary station forecasts; releases model weights, training code, and an ML-ready observational dataset (2007-2019) for building future end-to-end weather models (221+ stars, CC0-1.0)
 - [Earth-Copilot](https://github.com/microsoft/Earth-Copilot) - Microsoft's AI-powered geospatial Earth science application for natural-language exploration, visualization, and analysis of 130+ satellite collections, with STAC integration, multi-agent backend, MCP server, and deployable React/FastAPI stack (MIT, 2025)
+- [U-Cast (Rose-STL-Lab, ICML 2026)](https://github.com/Rose-STL-Lab/u-cast) - Surprisingly simple and efficient frontier probabilistic weather forecaster built on a standard U-Net trained with deterministic MAE pre-training followed by short CRPS fine-tuning via Monte Carlo Dropout, matching or exceeding the probabilistic skill of GenCast and IFS ENS at 1.5° resolution with >10× less training compute than leading CRPS models and >10× lower inference latency than diffusion models; trains in under 12 H200 GPU-days and generates a 15-day ensemble forecast in 3 seconds, with official code and pretrained checkpoints (UCLA, 41+ stars, Apache 2.0)
+- [OneForecast (ICML 2025)](https://github.com/YuanGao-YG/OneForecast) - Universal graph neural network framework for global-to-regional Earth system forecasting, combining multi-grid theory with a dynamic-system perspective to build multi-scale graphs that densify target regions for local high-frequency features; adaptive message passing with dynamic gating units is theoretically proven to act as high-pass filtering against over-smoothing, and a neural nested-grid method mitigates boundary information loss in high-resolution regional forecasts; extended to causally-coupled ocean-atmosphere cross-sphere modeling with strong extreme-event prediction, releasing inference/training code, pretrained weights, and preprocessed data (Renmin University & PolyU, 213+ stars, MIT License)
+- [WeatherGenerator (ECMWF, 2025-2026)](https://github.com/ecmwf/WeatherGenerator) - ECMWF's open-source machine-learning Earth system model developed by the WeatherGenerator Consortium with NVIDIA, trained on reanalyses, forecast data, and diverse observations across atmosphere, ocean, and land to provide a robust multi-scale model of Earth system dynamics; the first released version (v0.1, trained on ERA5) demonstrates global probabilistic forecasting skill on par with established AI models, with open training framework and config-driven multi-dataset ingestion pipeline (Apache 2.0)
 - [ClimaX](https://github.com/microsoft/ClimaX) - First foundation model for weather and climate by Microsoft, Vision Transformer-based architecture trained on heterogeneous datasets (ICML 2023)
 - [NeuralGCM](https://github.com/neuralgcm/neuralgcm) - Google Research's hybrid ML/physics atmospheric model combining learned dynamics with physical constraints, outperforming traditional models on 2-15 day forecasts and 40-year climate simulation, developed with ECMWF (Nature 2024)
 - [NVIDIA Earth-2](https://github.com/NVIDIA/earth2studio) - World's first fully open, accelerated weather AI software stack with Medium Range forecasting and Nowcasting models using generative AI (January 2026)
 - [Pangu-Weather](https://github.com/198808xc/Pangu-Weather) - Huawei's 3D high-resolution global weather forecast model at 0.25° resolution, first AI method to comprehensively outperform traditional NWP across all variables and lead times, integrated into ECMWF operational forecasts (Nature 2023)
 - [FuXi (Nature 2023)](https://github.com/tpys/FuXi) - Fudan University's cascade machine learning forecasting system for 15-day global weather prediction, employing a 3D Earth-specific transformer with hard-constraint techniques to achieve state-of-the-art accuracy against traditional NWP and AI baselines
 - [FengWu](https://github.com/OpenEarthLab/FengWu) - Shanghai AI Lab's deep learning-based global weather forecasting model pushing skillful forecasts beyond 10 days lead, with open-source inference code and pretrained ONNX model weights (arXiv 2023)
+- [Stormer (Argonne National Lab & U Chicago, NeurIPS 2024)](https://github.com/tung-nd/stormer) - Minimal-modification vision transformer for skillful and reliable medium-range weather forecasting, introducing weather-specific patch embedding, randomized dynamics forecasting over varying time intervals, and pressure-weighted loss; competitive at short range and outperforming prior methods beyond 7 days on WeatherBench 2 with orders-of-magnitude less training data and compute, with favorable scaling in model size and training tokens (MIT License)
 - [ai-models (ECMWF)](https://github.com/ecmwf-lab/ai-models) - ECMWF's unified framework and command-line tool to run AI-based weather forecasting models (GraphCast, Aurora, Pangu, NeuralGCM, FourCastNet) with operational ECMWF data infrastructure, enabling standardized inference and benchmarking across state-of-the-art meteorological AI systems (ECMWF, 576+ stars)
 - [Prithvi WxC](https://huggingface.co/ibm/prithvi-wxc) - IBM-NASA open-source 2.3B parameter weather and climate foundation model trained on 160 MERRA-2 variables, runs on desktop with fine-tuned variants for climate downscaling and gravity wave parameterization
 - [ClimateBench](https://github.com/duncanwp/ClimateBench) - Climate data benchmark for ML models
@@ -789,8 +838,13 @@ validated: false
 - [Earth-Agent](https://github.com/opendatalab/Earth-Agent) - LLM agent framework for Earth Observation with 104 specialized tools across 5 functional kits
 - [AI for Earth](https://planetarycomputer.microsoft.com/) - Microsoft's environmental AI
 
+#### Hydrology & Water Resources
+- [NeuralHydrology (KIT, HESS 2019/2021)](https://github.com/neuralhydrology/neuralhydrology) - Python library from KIT for training deep learning models on large-sample hydrology, introducing LSTM-based rainfall-runoff modeling that learns universal, regional, and local hydrological behaviors from hundreds of catchments; widely used for streamflow forecasting, flood prediction, and drought research, supporting the CAMELS and Caravan benchmark datasets (581+ stars, BSD-3-Clause, actively maintained)
+
 #### Geophysics & Seismology
 - [SeisBench](https://github.com/seisbench/seisbench) - A toolbox for machine learning in seismology, providing unified interfaces for deep learning seismic phase picking, earthquake detection, and waveform analysis across multiple benchmark datasets and pretrained models (397+ stars, actively maintained)
+- [EQTransformer (Nature Communications 2020)](https://github.com/smousavi05/EQTransformer) - Self-attention transformer performing simultaneous earthquake detection and P/S phase picking on continuous seismic waveforms, trained on the large-scale STEAD benchmark dataset and outperforming legacy detection methods with far fewer false positives; widely adopted by seismological observatories for earthquake monitoring, included in SeisBench model collections, and extended by efficient EQT-Mini/EQT-Lite successors for real-time deployment (Stanford, 418+ stars, MIT License, actively maintained)
+- [PhaseNet (GJI 2019)](https://github.com/AI4EPS/PhaseNet) - U-Net-style deep neural network for P/S seismic arrival-time picking trained on millions of waveforms from the Northern California Earthquake Data Center, achieving near-analyst picking precision at orders-of-magnitude higher speed and robustness to low signal-to-noise traces where STA/LTA fails; a foundational reference for deep-learning phase picking, integrated into SeisBench model collections and national seismic networks, with PhaseNet-DAS extending it to distributed acoustic sensing (Stanford AI4EPS, 386+ stars, MIT License, actively maintained)
 
 #### Remote Sensing & Geospatial AI
 - [TorchGeo](https://github.com/microsoft/torchgeo) - PyTorch domain library for geospatial deep learning providing standardized datasets, samplers, transforms, and pre-trained models for remote sensing, land cover mapping, and environmental monitoring (Microsoft, 4K+ stars)
@@ -820,6 +874,7 @@ validated: false
 - [BioCLIP 2 (NeurIPS 2025 Spotlight)](https://github.com/Imageomics/bioclip-2) - Biological vision foundation model trained on TreeOfLife-200M, yielding extraordinary accuracy on diverse biological visual tasks including habitat classification and trait prediction despite a narrow training objective (Ohio State University Imageomics Institute)
 - [Microsoft Biodiversity](https://github.com/microsoft/Biodiversity) - Microsoft AI for Good Lab's open-source biodiversity research hub providing AI models, edge devices, and tools for wildlife monitoring and conservation, including MegaDetector (camera trap animal detection), SPARROW (species recognition), PytorchWildlife (conservation AI toolkit), and bioacoustics analysis pipelines (1K+ stars)
 - [BirdNET-Analyzer](https://github.com/birdnet-team/BirdNET-Analyzer) - Deep learning-based bioacoustic monitoring framework for automated bird species identification from audio recordings, supporting 6,000+ species globally with real-time analysis, batch processing, and API deployment; foundational tool in biodiversity research, conservation biology, and ecological acoustic monitoring (Cornell Lab of Ornithology, 1.5K+ stars, MIT License)
+- [DeepForest (Methods in Ecology and Evolution 2020)](https://github.com/weecology/DeepForest) - First open-source deep learning package for detecting ecological objects — individual tree crowns and birds — in airborne RGB imagery, built on torchvision object detection with pretrained NEON tree-crown and bird models, simple custom-training workflows, and a pip-installable package; widely adopted for large-scale forest inventory, canopy mapping, and biodiversity surveys (Weecology Lab, University of Florida, 785+ stars, MIT License, actively maintained)
 
 ### 🧠 Social Sciences
 
@@ -834,6 +889,7 @@ validated: false
 ## 🏗 Engineering & Built Environment
 
 ### Mechanical, Aerospace & Industrial Engineering
+- [AgentFEM](https://github.com/haoming-luo/agentfem) - Open-source finite-element workflows connecting AI agents, simulation campaigns, and scientific learning datasets.
 - [Noether (Emmi AI)](https://github.com/Emmi-AI/noether) - Open software framework for Engineering AI built on transformer building blocks, enabling teams to build, train, and operate industrial simulation models across engineering verticals; includes ready-to-use recipes for CFD (AB-UPT on DrivAerML), external aerodynamics, and heat transfer (234+ stars, ENPL non-commercial license, 2026)
 
 ### Structural & Civil Engineering
@@ -847,6 +903,7 @@ validated: false
 
 ### Electrical & Electronics Engineering
 - [kicad-happy](https://github.com/aklofas/kicad-happy) - AI coding agent skills for KiCad electronics design that turn Claude Code, Codex, Gemini CLI, and other coding agents into full electronics design assistants; parses schematics and PCB layouts, builds power trees, audits connectors/ESD protection, validates passive networks, runs SPICE simulation, sources components from major distributors, and prepares boards for fabrication (aklofas, 974+ stars, MIT License, 2026)
+- [Circuit Training / AlphaChip (Google Research, Nature 2021)](https://github.com/google-research/circuit_training) - Open-source framework behind AlphaChip, the first ML system to design superhuman chip floorplans used in production Google TPU blocks; trains edge-based graph neural networks with distributed PPO and fine-tunes with simulated annealing to place macros under timing, congestion, and density constraints, generalizing across chip netlists (Google Research, 1.7K+ stars, Apache 2.0)
 
 ---
 
@@ -858,6 +915,8 @@ validated: false
 - [Llemma](https://github.com/EleutherAI/math-lm) - Open language model for mathematics (7B/34B) trained on Proof-Pile-2, outperforming Minerva at equal scale on MATH benchmark, with tool use and formal theorem proving in Lean without finetuning (EleutherAI, ICLR 2024)
 - [TimesFM (Google Research)](https://github.com/google-research/timesfm) - Pretrained time series foundation model for long-horizon forecasting across diverse scientific domains including climate variables, biomedical signals, and physical observations; decoder-only Transformer architecture with strong zero-shot generalization (19.8K+ stars, Apache 2.0, 2024-2025)
 - [Chronos (Amazon Science, NeurIPS 2024)](https://github.com/amazon-science/chronos-forecasting) - Pretrained time series foundation model for zero-shot forecasting across diverse scientific and real-world domains; tokenizes continuous time series into discrete bins to train transformer language models on large-scale corpora, achieving strong zero-shot generalization and competitive performance with task-specific supervised models on climate, energy, and health benchmarks (5.3K+ stars, Apache 2.0, 2024-2026)
+- [Moirai-MoE (Salesforce AI Research, NeurIPS 2024)](https://github.com/SalesforceAIResearch/uni2ts) - Universal time series forecasting via the UNI2TS library, training a single transformer with shared self-attention and specialized mixture-of-experts feed-forward blocks to achieve strong zero-shot generalization across heterogeneous domains including energy, weather, transportation, and health time series (1.6K+ stars, Apache 2.0, 2024-2026)
+- [Moment (CMU Auton Lab, ICML 2024)](https://github.com/moment-timeseries-foundation-model/moment) - Family of open time-series foundation models (MOMENT-1 small/base/large, T5-style encoder-only architecture with time series patching) pretrained on the curated Time Series Pile corpus, enabling zero/few-shot forecasting, classification, anomaly detection, imputation, and representation learning across scientific and real-world temporal data, with open weights on HuggingFace and pip-installable `momentfm` package (Auton Lab, CMU & UPenn, MIT License, 850+ stars)
 - [TabPFN (Prior Labs, Nature 2025)](https://github.com/PriorLabs/tabpfn) - Foundation model for tabular data that predicts on unseen real-world tables in a single forward pass, achieving accurate small-data classification and regression without task-specific training; widely applicable to scientific datasets with limited samples (7.4K+ stars, 2022-2026)
 - [TabFM (Google Research, 2026)](https://github.com/google-research/tabfm) - Scikit-learn compatible tabular foundation model for zero-shot classification and regression on mixed-type tabular datasets via in-context learning; applicable to diverse scientific datasets (1.8K+ stars, Apache 2.0)
 - [DeepInnovator (HKUDS, arXiv 2026)](https://github.com/HKUDS/DeepInnovator) - 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)
@@ -869,10 +928,12 @@ validated: false
 ### Domain-Specific Models
 - [ESM](https://github.com/facebookresearch/esm) - Protein language models
 - [BioNeMo Framework](https://github.com/NVIDIA/bionemo-framework) - NVIDIA's open-source platform for building and adapting biological AI models at scale, bundling ESM-2, Geneformer, MolMIM and DNA embedding models with recipes for single-GPU to multi-node training (2025)
+- [BioNeMo Recipes (NVIDIA)](https://github.com/NVIDIA-BioNeMo/bionemo-recipes) - TransformerEngine-accelerated checkpoints and training recipes for scaling biological foundation models (ESM-2, AMPLIFY, Geneformer, CodonFM) from single-GPU prototyping to multi-node FSDP training with FP8/MXFP8/NVFP4 precision, compatible with PyTorch, HF Accelerate, and PyTorch Lightning, plus sparse-autoencoder interpretability tools for biological foundation models (851+ stars, 2026)
 - [IBM FM4M](https://github.com/IBM/materials) - IBM's open foundation model family for materials and chemistry, covering SMILES, SELFIES, molecular graphs, 3D atom positions, and electron density grids, with a unified toolkit for representation learning and downstream prediction/generation (Apache 2.0, 2024-2025)
 - [ChemGPT](https://huggingface.co/ncfrey/ChemGPT-1.2B) - Chemistry-focused language model
 - [BioGPT](https://github.com/microsoft/BioGPT) - Biomedical text generation
 - [HuatuoGPT-o1 (2025)](https://github.com/FreedomIntelligence/HuatuoGPT-o1) - Open-source medical large language model for complex clinical reasoning, extending the o1 long-chain-of-thought paradigm to biomedical question answering and diagnostic inference (FreedomIntelligence, 1.3K+ stars)
+- [Med-Gemma (Google Health, 2025)](https://github.com/google-health/medgemma) - Google's open multimodal medical LLM family built on Gemma 3, including a 4B multimodal model handling medical images (radiology, pathology, dermatology) alongside text and a 27B text model for clinical reasoning; trained on de-identified medical data with checkpoints and inference code released under Apache 2.0 (1.6K+ stars)
 - [AntAngelMed (2026)](https://github.com/MedAIBase/AntAngelMed) - 103B-parameter open-source medical language model with 1/32 Mixture-of-Experts architecture, achieving HealthBench-leading performance among open-source models with only 6.1B active parameters; jointly developed by Ant Group and Zhejiang Province Health Information Center (MIT License)
 
 ---
@@ -895,6 +956,7 @@ validated: false
 
 ### Chemistry & Materials
 - [Materials Project](https://next-gen.materialsproject.org/) - Computational materials database
+- [MoleculeFinder Everyday Molecules](https://doi.org/10.5281/zenodo.23605767) - CC0 tables of 915 everyday molecules (PubChem computed properties, Wikidata names, US-government-sourced LD50s) matched by ingredient codes to 405 consumer products whose labels list them; versioned via a Zenodo concept DOI and refreshed weekly by a public ETL pipeline (2026)
 - [QM9](https://quantum-machine.org/datasets/) - Small molecule properties
 - [Open Catalyst Project](https://opencatalystproject.org/) - Catalyst discovery
 
@@ -944,7 +1006,10 @@ validated: false
 - [TorchMD](https://github.com/torchmd/torchmd) - 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)
 - [SO3LR](https://github.com/general-molecular-simulations/so3lr) - 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)
 - [TorchSim](https://github.com/TorchSim/torch-sim) - PyTorch-native atomistic simulation engine for the machine-learned interatomic potential (MLIP) era, enabling batched molecular dynamics and structural relaxation with automatic GPU memory management; supports MACE, Fairchem, SevenNet, ORB, MatterSim and other popular MLIPs with up to 100x speedup over ASE (Radical AI, AI for Science 2026, 468+ stars, MIT License)
+- [GPUMD](https://github.com/brucefan1983/GPUMD) - High-performance, GPU-native molecular dynamics package supporting empirical potentials and neuroevolution potentials (NEPs) with a built-in `nep` training executable, achieving DFT-level accuracy at orders-of-magnitude lower training cost than competing MLIPs; widely adopted for heat transport (NEP/HNEP), lattice dynamics, mechanical properties, shock compression, and path-integral MD, with CUDA and ROCm support and extensive tutorials (849+ stars, GPL-3.0, Comput. Phys. Commun. 2017 / Phys. Rev. B 2021, actively maintained)
 - [Newton](https://github.com/newton-physics/newton) - GPU-accelerated differentiable physics simulation engine built on NVIDIA Warp, supporting rigid/soft body, cloth, and gradient-based optimization for scientific ML, initiated by Disney Research, DeepMind, and NVIDIA (Linux Foundation, Apache 2.0, 2025)
+- [Genesis World (Genesis-Embodied-AI, Dec 2024+)](https://github.com/Genesis-Embodied-AI/Genesis) - Universal generative physics simulation platform integrating rigid body, FEM, MPM, particle (PBD/SPH), and cloth solvers with explicit multi-solver coupling in a single differentiable scene, GPU-graph-accelerated massively parallel simulation, a photorealistic renderer (Nyx), and a cross-platform compiler (CUDA/ROCm/Metal/Vulkan/CPU); scales from laptops to datacenter GPUs for physics simulation, synthetic data generation, and physical-AI/robotics research (30K+ stars, Apache 2.0, actively maintained, now officially supported by Genesis AI)
+- [NVIDIA Warp](https://github.com/NVIDIA/warp) - Python framework for writing high-performance GPU simulation and graphics kernels with first-class automatic differentiation, enabling differentiable physics, molecular dynamics, soft-body and cloth simulation, robotics, and CFD adjoints compiled to CUDA; serves as the underlying engine for differentiable simulation projects like Newton and integrates with PyTorch, JAX, and OpenUSD (NVIDIA, 7.1K+ stars, Apache 2.0)
 - [JAX-CFD](https://github.com/google/jax-cfd) - Computational fluid dynamics in JAX, enabling differentiable Navier-Stokes simulations with automatic differentiation for ML-accelerated CFD research, supporting turbulence modeling, convection-diffusion, and complex boundary conditions on CPUs and GPUs (Google Research, 947+ stars)
 - [PennyLane](https://github.com/PennyLaneAI/pennylane) - Cross-platform library for differentiable programming of quantum computers with automatic differentiation, enabling hybrid quantum-classical machine learning for quantum chemistry, quantum physics, and NISQ algorithm research (Xanadu, 3k+ stars)
 - [Qiskit](https://github.com/Qiskit/qiskit) - Open-source SDK for working with quantum computers at the level of extended quantum circuits, operators, and primitives, enabling quantum algorithm development for quantum chemistry, materials science, and optimization research (IBM, 7.4K+ stars, Apache 2.0)