Biological models

318 model records in release 2026-10-23-5347695500ed (release manifest, gzip). Showing 169–192; page 8 of 14.

Evaluated configurations, methods and pipelines are listed separately. Names alone do not establish equivalent models or checkpoints.

• LucaVirus

  Microbes and communities · DNA and genomes

  LucaVirus unified nucleotide-protein language model for viruses.

• MADGEN

  Metabolomics

  MADGEN mass-spectrum-guided de novo molecule generation.

• MassFormer

  Metabolomics

  MassFormer graph transformer spectrum predictor, used as the predictor head in FlexMS.

• MassSpecGym DeepSets fingerprint predictor

  Metabolomics

  DeepSets fingerprint prediction baseline from MassSpecGym that predicts a molecular fingerprint from an MS/MS spectrum to rank candidate molecules (Fourier-feature variant in MSAlign).

• METAGENE-1

  Microbes and communities

  METAGENE-1 is an autoregressive DNA/RNA sequence model trained on wastewater metagenomic data.

• MetageNN

  Microbes and communities

  Feed-forward neural network on 6-mer profiles of 1 kbp sequences sampled from reference genomes; long reads are split into 1 kbp chunks and classified by majority vote at genus level.

• MFE

  Molecular interactions

  MFE: multimodal extraction model combining 3D CNN, GVP-GNN and ProtTrans features.

• MIF

  Proteins and complexes

  Masked inverse folding model, a structured graph neural network that reconstructs corrupted protein sequences conditioned on the backbone structure, scored in ProteinGym as a zero-shot structure-based substitution baseline.

• MIF-ST

  Proteins and complexes

  Masked inverse folding model with sequence transfer, which feeds the outputs of a pretrained sequence-only protein masked language model into a structure-conditioned graph neural network, scored in ProteinGym as a zero-shot substitution baseline.

• MIMIC

  RNA and transcriptomes · Proteins and complexes

  MIMIC represents DNA, RNA, protein and associated molecular measurements in a shared multimodal model.

• MIST

  Metabolomics

  Spectrum-to-molecule retrieval model that pretrains on fingerprint prediction, fine-tunes with contrastive learning and augments its data with a forward model that simulates spectra; it needs the target's molecular formula.

• Mistral-DNA

  DNA and genomes

  Mistral-architecture DNA language model.

• MixingDTA

  Molecular interactions

  MixingDTA: meta-ensemble affinity model using MolFormer and ESM3 features.

• MochiBind

  Molecular interactions · Proteins and complexes

  MochiBind: sequence-only pairwise antibody-antigen binding predictor using ESM-2 (esm2_t36_3B_UR50D) embeddings and a pairwise scoring module.

• ModernGENA

  DNA and genomes

  BERT-style DNA language model built on the ModernBERT backbone with a 32k BPE vocabulary.

• MolForge

  Metabolomics

  MolForge fingerprint-to-molecule decoder, used with a MIST spectrum encoder.

• MolMS

  Metabolomics

  Point-cloud molecular spectrum predictor, used as a predictor head in FlexMS.

• MP-RNA

  Microbes and communities · DNA and genomes

  MP-RNA RNA language model.

  Record: lit-microbial-20261011-virobench-model-mp-rna

• MP-RNA

  RNA and transcriptomes

  MP-RNA RNA language model.

  Record: lit-rna-20261011-rnalm-zeroshot-model-mp-rna

• MPNN

  Protein function

  Message passing neural network over the protein graph.

• mRNA-FM

  RNA and transcriptomes

  mRNA-FM encodes coding RNA with codon-level tokens to produce representations for downstream analysis.

• MS-BART

  Metabolomics

  MS-BART spectrum-to-molecule generation model.

• MSA Transformer

  Proteins and complexes

  Protein language model that takes a multiple sequence alignment as input, interleaving row and column attention, trained by masked language modelling across many protein families; ProteinGym scores it as a zero-shot alignment-based substitution baseline.

• MSAlign

  Metabolomics

  MSAlign retrieves candidate molecules from tandem mass spectra by aligning pretrained molecular and spectral representations.

This index reflects a dated catalogue, not an exhaustive census. Source checking does not mean independent reproduction; compare results only under compatible protocols, datasets and metrics.