Loading…
Every biological foundation model, evaluated and ranked by the bio.rodeo team
Showing 1–24 of 2565 models
Single-cell foundation model pretrained with cell-annotation and gene-regulon supervision on 179,067 profiles, then transferred as a frozen encoder.
T cell receptor foundation model that scores peptide-MHC-TCR recognition and generates pMHC-conditioned CDR3β candidates from one shared backbone.
Anticancer peptide prediction specialized to breast cancer, running an MLP head over frozen ProtBERT embeddings of short peptide sequences.
All-atom de novo protein design that denoises sequence and structure as a single Cartesian process, reading residue identity from atomic occupancy.
3D small-molecule generation conditioned on an interaction profile of shape, electrostatics and directional pharmacophores rather than a pocket.
Plasma proteomics foundation model for prospective disease risk prediction across 144 diseases. Unmeasured proteins are omitted rather than imputed.
Diploid genomic language model that writes both homologues as one token stream, so zygosity, allele dosage and cis-trans phase reach the encoder.
Peptide-HLA class I binding predictor that deconvolves multi-allele immunopeptidomics, pairing frozen ESM-2 embeddings with attention-based MIL.
Generative EHR foundation model that forecasts which clinical events come next and when, then adapts to pan-cancer risk stratification.
EEG model family that distills a SimCLR-pretrained SE-ResNet teacher into edge-sized students for abnormality, motor-imagery and emotion decoding.
Antibody language model that reads all four frameworks as a prompt and generates CDR1-CDR3 jointly, with weighted-DPO alignment to design rewards.
Self-supervised 4D live-cell microscopy model embedding mitochondrial shape and motion. Frozen features predict membrane potential at R² = 0.91.
Phylogeny-informed transformer that detects episodic diversifying selection in codon alignments, amortizing maximum-likelihood dN/dS tests.
Protein language model that grows its own depth and width during masked-language pretraining on the human proteome, reaching 76.0% Pfam-50 accuracy.
De novo small-molecule generation conditioned on disease-reversal gene expression, with a contrastive transcriptome encoder steering a MolT5 decoder.
Antibody-antigen binding affinity predictor that scores the interface as a graph over frozen ESM-2 embeddings, jointly regressing log10(KD) and ΔG.
Coding potential classifier for assembled transcripts, joining pretrained RNA transformer embeddings with ORF and UTR descriptors in a random forest.
Label-free virtual staining framework that turns gigapixel spectral whole-slide images into H&E histology without tiling seams or color drift.
Self-supervised ViT-Small/14 domain-adapted from DINOv2 to chest radiographs, linear-probed for stroke and heart failure detection.
Shared MHC class II neoantigen design for breast cancer, pairing retrieval-augmented ProtT5 embeddings with an MCNN-BiLSTM immunogenicity classifier.
Intraoperative X-ray to CT registration that reads the preoperative volume as a second input, removing the per-patient training step entirely.
RNA 3D structure evaluator that ranks candidate conformations without a reference. Sequence conditioning comes from a frozen RNA language model.