Find open-source science resources

PertMind is a biological language model built around a central discovery: public cellular perturbation atlases can be reorganized into reinforcement-learning environments, where measured gene responses act as computable reward signals for biological reasoning.

A from-scratch, decoder-only protein language model for the phosphotransferase superfamily (EC 2.7.-: protein kinases plus sugar/lipid/nucleotide kinases), trained entirely locally on Apple Silicon via MLX — no cloud compute, no fine-tuning of an existing model.

Run locally - Benchmarks - Whitepaper - Model details - Responsible use

Native OpenMed and OpenMedKit vision-language inference for Apple Silicon, including local clinical-document and chart workflows on Mac, iPhone, and iPad.

A 350M encoder that finds nine types of personally identifiable information across 17 languages and returns exact character spans for review and redaction.

UBio-MolFM is a foundation-model suite for molecular modeling, designed for bio-systems. This release, UBio-MolFM-V1.5 (Stage 3), is built on the E2Former-V2 linear-scaling equivariant transformer and is the checkpoint used for every simulation reported in UBio-MolFM: Enabling Biomolecular Dynamics…

BondShift: Organic Mechanism Reasoning

Vision-language model for dermatology, pretrained with MAGEN (Multi-Agent data GENeration) and O-MAKE (Ontology-based Multi-Aspect Knowledge-Enhanced pretraining).

A clinical reasoning assistant for early-stage Alzheimer's assessment. It joins a 3D MRI + biomarker classifier (Vbai-2.6AD) to a reasoning LLM (Gemma 4 12B) inside a single forward pass — the diagnosis is passed as vectors, not text.

Contrastively fine-tuned ESM-C 300M producing fixed-length protein embeddings where biological similarity maps to embedding proximity. Intended for retrieval, clustering, and nearest-neighbour transfer.

SERAPH is a deep learning model designed for 3-state (Q3) protein secondary structure prediction. It processes raw single amino acid sequences and predicts residue-level secondary structure states: Alpha Helix (H), Beta Sheet (E), or Coil/Loop (C).

!TVBP Architecture !Parameters !Trainable !Brownian Reservoir !Framework !Biology

Technical Report 🧬

Nesso-1 is a fast, structure-based protein–ligand binding-affinity model. Given a protein sequence and a ligand (SMILES / CCD code / SDF), it predicts a binding affinity scalar along with a binder/non-binder score.

This 1,120,772,224-parameter nucleotide-level causal language model is a member of the eight-model MarinDNA v0.5 parameter-scaling ladder developed with Marin. This repository contains only the final step-215573 checkpoint from run dna-bolinas-scaling-v0.5-h1920-p1B-0dc6f4, with its tokenizer…

MarinDNA m5.1 is a 1.12B-parameter, nucleotide-level causal language model developed with Marin. This is the final m5.1 base-model checkpoint at step 59,158 from run dna-bolinas-mix-v0.9-p1B-i24-exp135-zoonomia-m5.1-bef41e, released with the A 1B standard Transformer rivals Evo 2 40B on variant…

For a convenient overview and download list, visit our model page for this model.

💻 GitHub Repo • 🤗 Model Collections • 📖 MemSFT Paper

!Format !Task !Params !Type !License

HealthGPT-LoRA is a biomedical question-answering model built by fine-tuning Meta Llama 3.2 3B Instruct using QLoRA (PEFT) on the PubMedQA dataset.

ProtSent-V2 35M plus one more contrastive pass on a fresh draw of the corpus, with a DMS/ProteinGym CoSENT target and a Global Orthogonal Regularization term added.

A DINOv2 ViT-S/14-reg fine-tuned so that an image of a molecular structure diagram embeds where its molecule embeds in the frozen MIST-28M embedding space. Objective: smooth-L1 regression onto the frozen target, no negatives (the JEPA move).

A DINOv2 ViT-S/14-reg fine-tuned so that an image of a molecular structure diagram embeds where its molecule embeds in the frozen MIST-28M embedding space. Objective: SigLIP sigmoid pairwise loss.

Contrastively fine-tuned ESM-2 150M producing fixed-length protein embeddings where biological similarity maps to embedding proximity. Intended for retrieval, clustering, and nearest-neighbour transfer.

🩺 HealthGPT-Pro: A High-Performance Multimodal Large Language Model for Medical Understanding and Analysis

A binary healthy hard coral vs bleached hard coral classifier built on top of the ReefNet species LoRA model BobDerBaum/bioclip-2.5-vith14-reefnet-lora, which provides the fine-tuned vision-encoder LoRA adapters. Only a small linear head is trained on top (frozen backbone + LoRA + 2-way linear…

A LoRA fine-tune of imageomics/bioclip-2.5-vith14 trained contrastively on the ReefNet 1.0 coral-reef species dataset (ReefNet/ReefNet-1.0), 92-class global curated split (train 48,312 / image-val 32,792 / image-test 33,090 / source-val 8,074; split cache 56ea94e36f9f).

A lightweight language model designed to answer science questions clearly and accurately in English.

!Nucleus Resynthesis

For a convenient overview and download list, visit our model page for this model.

Winnow recalibrates confidence scores and provides FDR control for de novo peptide sequencing (DNS) workflows. This repository hosts a calibrator trained on the HeLa Single Shot dataset as referenced in our paper: De novo peptide sequencing rescoring and FDR estimation with Winnow.

Winnow recalibrates confidence scores and provides FDR control for de novo peptide sequencing (DNS) workflows. This repository hosts a pretrained, general-purpose calibrator that maps raw InstaNovo model confidences and complementary features (mass error, retention time, beam features, fragment…

ESMFold2 is a state-of-the-art model for protein structure prediction and design that defines a new frontier for speed and accuracy. The model predicts high-resolution, all-atom 3D protein structures directly from amino acid sequences, with optional multiple sequence alignment (MSA) input for…

ESMFold2 is a state-of-the-art model for protein structure prediction and design that defines a new frontier for speed and accuracy. The model predicts high-resolution, all-atom 3D protein structures directly from amino acid sequences, with optional multiple sequence alignment (MSA) input for…

464 fine-tuned DNABERT models for regulatory variant effect prediction

GB.DNA-7B is DNA foundation model trained on 10.6 billion nucleotides from 796 species, enabling genome mining, in silico mutagenesis studies, gene expression prediction, and directed sequence generation.