Biological models

320 model records in release 2026-11-02-a7f03c9e55a5 (release manifest, gzip). Showing 1–24; page 1 of 14.

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

• 3UTRBERT

  RNA and transcriptomes

  3UTRBERT, a BERT language model pretrained on 3' UTR sequences.

• Agro Nucleotide Transformer

  DNA and genomes

  AgroNT learns DNA representations from plant reference genomes for plant molecular prediction tasks.

• AIDO Protein-RAG

  Proteins and complexes

  Retrieval-augmented protein language model from the AIDO.Protein-RAG work, which combines a pretrained protein language model with retrieved multiple sequence alignments; ProteinGym lists AIDO with alignment and structure inputs and scores it as a zero-shot substitution baseline.

• AIDO.DNA

  DNA and genomes

  AIDO.DNA genomic language model.

• AIDO.RNA

  Microbes and communities · DNA and genomes

  AIDO.RNA RNA language model family.

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

• AIDO.RNA

  RNA and transcriptomes

  AIDO.RNA RNA foundation model (650M and 1.6B parameter checkpoints).

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

• Akita

  DNA and genomes

  Akita contact-map model.

• AlphaFold 2

  Protein structure · Proteins and complexes

  Deep-learning protein structure prediction model evaluated in this comparison.

• AlphaFold 3

  Protein structure

  AlphaFold 3 predicts three-dimensional structures of complexes containing proteins, nucleic acids and other molecular components. It combines a Pairformer representation network with an atomic-coordinate diffusion model. This entry describes the model and local implementation; the hosted AlphaFold Server has a separate profile.

• AlphaGenome

  DNA and genomes

  AlphaGenome predicts regulatory activity and variant effects from long DNA sequences, with outputs for expression, splicing, chromatin and contact maps.

• AlphaPeptDeep

  Proteins and complexes

  AlphaPeptDeep modular Transformer fragment intensity predictor (about 1.8M parameters).

• AttentiveFP

  Protein function

  Attentive fingerprint graph network.

• BACPI

  Molecular interactions

  Sequence-based affinity predictor whose 1D convolutional protein network and ligand graph attention transformer are joined by bi-directional attention.

• Basenji

  DNA and genomes

  Basenji predicts quantitative regulatory activity along DNA and scores the effects of sequence changes.

• BERTax

  Microbes and communities

  BERT encoder pretrained on 1,500-nt genome fragments (3-mer tokens) and fine-tuned to predict superkingdom, phylum and, in the final model, genus.

• BioFM

  DNA and genomes

  Transformer decoder genomic model whose BioToken tokeniser encodes variants and annotated regions alongside sequence, used to predict expression, splicing and variant pathogenicity.

• BiRNA-BERT

  Microbes and communities · DNA and genomes

  BiRNA-BERT RNA language model.

  Record: lit-microbial-20261011-virobench-model-birna-bert

• BiRNA-BERT

  RNA and transcriptomes

  BiRNA-BERT RNA language model with adaptive nucleotide and BPE tokenisation.

  Record: lit-rna-20261011-rnalm-zeroshot-model-birna-bert

• Boltz

  Molecular interactions

  Boltz predicts biomolecular complex structures; Boltz-2 also predicts binding affinity.

• Boltz-2

  Molecular interactions

  Boltz predicts biomolecular complex structures; Boltz-2 also predicts binding affinity.

• Borzoi

  DNA and genomes

  Model trained to predict RNA-seq coverage from DNA sequence.

• BPNet

  DNA and genomes

  BPNet convolutional architecture trained from scratch per task.

• Caduceus

  DNA and genomes

  Self-supervised DNA language model pretrained on human genomic sequence.

• Carbon-X

  DNA and genomes

  Carbon-X reads raw DNA inside a chat conversation and answers in English, predicting variant pathogenicity, classifying sequences and predicting the next bases.

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.