AI tools for biology, verified.
188 hand-checked tools across 15 categories, from single-cell, spatial omics, protein design, genomics, drug discovery, agents, cloud labs, imaging, datasets, sequencing, proteomics/metabolomics, CRISPR, RNA, microbiome and clinical LLMs.
Recently released Oct 2 – Oct 16
Language-conditioned model that predicts cell-level spatial gene expression from histology across tissues and gene panels.
Sequence-only pLM ensemble predicting membrane-protein interaction interfaces; proteome-scale atlas covering 80k+ reviewed TM proteins.
Text-to-structure all-atom protein design: generate monomers and binders from natural-language prompts.
Illumina's successor to SpliceAI: predicts splice-site usage and full transcript isoforms from sequence.
1.2B DNA model that annotates protein-coding genes across eukaryotes; ships a 48K-assembly annotation database.
Conversational ReAct agent for gene-set enrichment: 220+ libraries, Europe PMC literature grounding, interactive plots and human-in-the-loop refinement.
Virtual-cell model using curated biological knowledge to predict perturbation responses in unseen contexts.
Google DeepMind diffusion pipeline for de novo enzyme design via catalytic motif scaffolding.
All tools
Language-conditioned model that predicts cell-level spatial gene expression from histology across tissues and gene panels.
Sequence-only pLM ensemble predicting membrane-protein interaction interfaces; proteome-scale atlas covering 80k+ reviewed TM proteins.
Text-to-structure all-atom protein design: generate monomers and binders from natural-language prompts.
1.2B DNA model that annotates protein-coding genes across eukaryotes; ships a 48K-assembly annotation database.
Illumina's successor to SpliceAI: predicts splice-site usage and full transcript isoforms from sequence.
Conversational ReAct agent for gene-set enrichment: 220+ libraries, Europe PMC literature grounding, interactive plots and human-in-the-loop refinement.
Google DeepMind diffusion pipeline for de novo enzyme design via catalytic motif scaffolding.
Virtual-cell model using curated biological knowledge to predict perturbation responses in unseen contexts.
Open generative model for designing protein and peptide binders against diverse targets.
NVIDIA latent flow model for joint atomistic protein sequence-structure generation.
DeepMind model predicting regulatory effects of DNA variants at base resolution.
General-purpose biomedical AI agent with 150+ tools and databases.
Arc Institute virtual-cell model predicting transcriptomic responses to perturbations across cell contexts.
Google's open Gemma 3 variants for medical text and image comprehension.
FutureHouse multi-agent system automating hypothesis generation and data analysis for drug discovery.
CZI cross-species generative single-cell foundation model trained on 112M cells from 12 species.
Atom-level enzyme active-site scaffolding with diffusion, no sequence-position constraints.
NVIDIA flow-matching model for large-scale protein backbone generation.
100M-cell single-cell drug perturbation atlas.
Genome language model across all domains of life with 1M-token context.
Trainable PyTorch reproduction of AlphaFold 3 by ByteDance.
Generative model emulating protein equilibrium structural ensembles.
RNA language-model-based method for 3D RNA structure prediction.
Foundation model for joint segmentation, detection and recognition across nine biomedical imaging modalities.
Open biomolecular interaction models (Boltz-1, Boltz-2) for complex structure and binding-affinity prediction.
One-shot de novo binder design pipeline with AF2 hallucination.
High-accuracy RAG agent over scientific papers.
Multimodal foundation model for molecular structure prediction.
Gymnasium for training language agents on scientific tasks (molecular cloning, literature QA).
Multimodal generative protein language model over sequence, structure and function (ESM3; also ESM C embeddings).
100M-parameter foundation model trained on 50M human single-cell transcriptomes.
LLM agent that designs genetic perturbation experiments (e.g. CRISPR screens).
Whole-slide digital pathology foundation model.
Structure prediction of proteins, nucleic acids, ligands and their complexes; free web server and code, weights on request for non-commercial use.
AI-designed Cas9-like gene editor from Profluent's language models, released openly.
Foundation model pretrained on dissociated and spatial single-cell data.
Vision-language foundation model for histopathology.
General-purpose self-supervised foundation model for computational pathology.
Structure-based de novo antibody and nanobody design with fine-tuned RFdiffusion, ProteinMPNN and RF2.
Bidirectional, reverse-complement-equivariant DNA language model.
7B genomic foundation model (StripedHyena) for prokaryotic DNA, RNA and protein design at single-nucleotide resolution.
RNA foundation model trained on Ribonanza chemical-mapping data for structure and reactivity.
Open 7B medical LLM based on Mistral, further pretrained on PubMed Central.
Paired-chain antibody language model trained to reduce germline bias.
GPT-4-driven agent that plans and executes chemistry experiments on lab hardware.
Benchmarks for protein fitness and mutation effect prediction.
Self-supervised transformer foundation model for tandem mass spectra of small molecules.
Cross-species zero-shot universal cell embedding foundation model.
Open medical LLMs (7B/70B) adapted from Llama 2 on curated medical corpora.
AstraZeneca's generative molecular design framework.
Genomic language model over multi-species alignments for genome-wide variant effect prediction.
Generalized biomolecular modeling for proteins, nucleic acids, small molecules.
Proteome-wide pathogenicity predictions for 71M missense variants (repo archived; precomputed scores available).
Predicts RNA-seq coverage from DNA sequence.
Transformer and diffusion models for de novo peptide sequencing from tandem mass spectra.
Graph-based prediction of transcriptional outcomes of genetic perturbations.
Plausibility checks and benchmark for docking/structure poses.
Hardware-agnostic Python interface for liquid-handling robots and lab automation.
Long-range genomic foundation model at single-nucleotide resolution.
Framework for LLM-based conversational biomedical research.
LLM chemistry agent augmented with 18 expert tools for synthesis and molecule tasks.
Benchmarking platform for ML in drug discovery.
Biomedical vision-language assistant trained on PubMed Central figure-caption data.
Transformer pretrained on ~30M single-cell transcriptomes for network biology predictions.
Deep learning structure prediction for antibodies, nanobodies and TCRs (ABodyBuilder2 etc.).
Geometric deep learning for 3D RNA inverse design (sequence design from backbone).
Multi-omics single-cell optimal transport for space and time mapping.
Missense variant pathogenicity model trained on primate population variation and 3D protein structure.
Generative pretrained transformer foundation model for single-cell multi-omics.
Unified data framework for spatial omics technologies.
Fast antibody structure prediction using AntiBERTy embeddings and graph networks.
Feature-free DIA proteomics search engine with transfer learning from AlphaPeptDeep.
Marker-gene-based taxonomic profiling of metagenomes, including uncharacterized species.
Living benchmarks for single-cell analysis tasks.
Bias-corrected base-resolution deep learning models of ATAC/DNase-seq chromatin accessibility and TF motifs.
Attention-based model predicting prime editing efficiency to design pegRNAs.
DNA foundation models pretrained on 3,200+ human and 850 species genomes.
Self-supervised deep learning metagenomic binning for short and long reads.
Curated single-cell data portal and Census API with 100M+ cells.
Diffusion model for de novo protein backbone design and binder design.
Oxford Nanopore's production basecaller: transformer/LSTM models turning raw nanopore signal into reads, with modified-base calling and duplex.
Diffusion generative model for molecular docking.
Transformer for de novo peptide sequencing directly from MS/MS spectra.
Neural-network and marker-based identification of viruses and plasmids in sequencing data.
Graph self-supervised contrastive learning for spatial transcriptomics.
Protein language models and single-sequence structure prediction (Meta repo archived; models still widely used via Hugging Face).
Machine-learning estimation of completeness and contamination of metagenome-assembled genomes.
Deep learning framework predicting peptide MS2 spectra, retention time and CCS for proteomics.
Protein language models (up to 6.4B) for sequence generation and fitness prediction.
Broad GPP model for SpCas9 sgRNA on-target activity, used in CRISPick.
Deep-learning inverse folding: designs sequences for given backbones.
Fast structural-variant caller for long reads with population-scale and mosaic SV detection.
RNA foundation model pretrained on 23M non-coding RNAs for structure and function tasks.
Cloud platform for bioinformatics workflows and data.
Fast and accurate protein structure search.
Spatial single-cell analysis toolkit in the scverse ecosystem.
Antibody language model for heavy/light chains, used for residue restoration and embeddings.
Transformer that corrects PacBio HiFi consensus reads, increasing yield and accuracy.
BERT-style antibody language model trained on 558M OAS sequences.
Models for detecting modified bases (e.g. 5mC, 6mA) from nanopore signal.
Trainable open-source reproduction of AlphaFold2 in PyTorch.
PyTorch platform for drug discovery ML.
Sequence-based deep learning model for single-cell ATAC-seq representation and TF activity.
Graph attention autoencoder for spatial domain identification.
Antibody humanization (Sapiens) and humanness evaluation (OASis) platform.
Predicts 21,907 chromatin profiles from sequence and maps variants to regulatory sequence classes.
Fast, accessible protein folding with MMseqs2 + AlphaFold in Colab.
DeepMind's open-source protein structure prediction model (monomer and multimer).
Multimodal omics framework for single-cell multi-omics in Python.
Deep-learning germline small-variant caller for long reads (ONT, PacBio) and short reads.
Automated cell type annotation with logistic regression models trained on curated atlases.
Predicts gene expression from sequence with long-range interactions.
Optimal-transport alignment and integration of spatial transcriptomics slices.
Fast, widely used open-source molecular docking engine.
ML-ready datasets and benchmarks for therapeutics.
Siamese neural network predicting structural similarity between MS/MS spectra for metabolomics.
Unsupervised evolutionary VAE for predicting pathogenicity of missense variants.
In silico gene perturbation via GRN inference from single-cell multi-omics.
Functional profiling of metagenomes and metatranscriptomes (gene families and pathways).
Cell fate mapping using Markov chains on single-cell data.
Bayesian mapping of fine-grained cell types in spatial transcriptomics.
Haplotype-aware variant calling pipeline for nanopore reads combining PEPPER, Margin and DeepVariant.
Probabilistic deep generative models for single-cell omics (scVI, totalVI, scANVI...).
Bayesian clustering and resolution enhancement of spatial transcriptomics.
Maps single-cell profiles onto spatial data with deep learning.
Fast image-like U-Net for RNA secondary structure prediction.
RNA secondary structure model trained via multitask learning on Eterna player data.
Python library for importing, cleaning and comparing mass spectra for metabolomics.
Robust cell type decomposition for spatial transcriptomics.
Scalable single-cell chromatin accessibility (scATAC-seq) analysis in R.
Oxford Nanopore's research basecaller for training and testing new basecalling models.
Generalist deep learning model for cellular segmentation.
Deep learning retention-time predictor that generalizes to unseen peptide modifications.
Fast, interactive multi-dimensional image viewer for Python.
Deep learning RNA secondary structure prediction including non-canonical pairs and pseudoknots.
RNA velocity estimation with dynamical modeling.
Base-resolution neural network for TF binding profiles (ChIP-nexus/ChIP-seq) and motif syntax discovery.
Extensible platform for microbiome (amplicon and shotgun) data science.
Web tool for designing CRISPR-Cas9, Cas12 and TALEN targets with efficiency predictions.
Metabolomics software for molecular formula and structure annotation from MS/MS with CSI:FingerID and CANOPUS.
Message passing neural networks for molecular property prediction.
Predicts peptide fragmentation spectra (and with DeepLC/IM2Deep) for spectral libraries and rescoring.
Deep learning prediction of splicing effects from pre-mRNA sequence (repo archived).
Variational autoencoder for metagenomic binning.
Fast, sensitive batch integration of single-cell data.
Analysis of genome-editing outcomes from amplicon sequencing, including base and prime editing.
PyTorch library for training and applying sequence-level deep learning models (DeepSEA-style).
Fast k-mer based taxonomic classification of metagenomic reads.
Star-convex object detection for nuclei segmentation.
Neural-network-powered data-independent acquisition (DIA) proteomics processing software.
Assigns GTDB taxonomy to bacterial and archaeal genomes and MAGs.
Neural-network consensus polishing and variant calling for nanopore assemblies.
Scalable Python toolkit for single-cell gene expression analysis.
Sequence-to-function CNNs predicting chromatin accessibility, histone marks and expression tracks from DNA.
Google deep learning variant caller for Illumina, PacBio HiFi and ONT data.
Open-source software for digital pathology and whole slide image analysis.
Web tool for CRISPR guide selection with on-target efficiency and off-target scoring for 500+ genomes.
Deep learning toolkit for drug discovery, materials and biology.
R toolkit for single-cell genomics, integration and multimodal analysis.
High-performance molecular dynamics toolkit with GPU support.
Standard RNA secondary structure prediction and design package (RNAfold, RNAinverse).
Public single-cell and spatial (Visium, Xenium) example datasets.
Cloud lab for protein expression and binding characterization, binder design competitions.
200M+ predicted protein structures from DeepMind and EMBL-EBI.
R&D cloud with ELN, molecular biology and AI tooling.
Critical Assessment of protein Structure Prediction community experiment.
Image analysis software for high-throughput cell phenotyping.
AI search engine over peer-reviewed research.
AI scientist agents (formerly FutureHouse Crow, Falcon, Owl, Phoenix) for literature search and chemistry.
AI research assistant for literature search and data extraction.
Remote-controlled cloud laboratory running 200+ instrument types.
Cell programming platform and automated lab services.
Comprehensive R toolbox for spatial data analysis and visualization.
Hub listing of biology-tagged models.
Data portal of the Human Cell Atlas consortium.
Open-source platform for LC-MS and imaging-MS data processing in metabolomics.
NVIDIA platform and NIMs for training and deploying biology foundation models.
Open-source liquid-handling robots with a Python Protocol API and AI protocol generation.
Archive of experimentally determined 3D biomolecular structures.
Open-source cheminformatics and machine learning toolkit.
Cloud computational chemistry platform with ML-accelerated simulations.
Marketplace for outsourced R&D and CRO services.
Robotic cloud lab automation for life sciences.
No-code web platform running 100+ protein and molecular ML models.
Synthetic DNA, gene fragments and antibody discovery services.
Comprehensive resource of protein sequence and function.