ContrasTED: contrastive domain embeddings for scalable remote homology classification
- 1University College London
Abstract
Protein structure prediction has expanded structural databases to hundreds of millions of domains. Classifying these domains into homologous superfamilies reveals evolutionary and functional relationships that can persist despite low sequence similarity. As the size of structural databases continues to grow, homology classification requires methods that combine scalability with accuracy. Here we present ContrasTED, which uses CATH-supervised center-contrastive learning to project structure-aware embeddings into a domain-level metric space for nearest-centroid superfamily assignment. On a sequence-filtered S20 benchmark (n = 1,028), superfamily assignment accuracy reached 92.9% (1-NN) and 91.4% (nearest centroid), exceeding sequence search, profile HMMs, Foldseek, and a classifier trained on embeddings. The learned latent space separates superfamilies while retaining structural information below 20% sequence identity, with the largest gains among sparsely represented superfamilies. ContrasTED produces 4.67 million new candidate assignments across 3,796 superfamilies in The Encyclopedia of Domains (TED), extending annotation coverage beyond previous structure-based methods.
Related
- Contrastive Neural Embeddings Reveal Individual Traits Beyond Conversational Role
- Interrogating contrastive learning embeddings for structure-based virtual screening: a case study on DrugCLIP
- Multicriteria assessment of cropping systems under contrasted soil fertility managements: evidence from mixed farming in the Madagascar highlands
- Winter warming and agricultural plant diversification promote ground beetles and spiders with contrasted functional thermal responses
- Structural comparison of homologous protein-RNA interfaces reveals widespread overall conservation contrasted with versatility in polar contacts
BibTeX
@misc{miller2026contrasted,
title = {{ContrasTED: contrastive domain embeddings for scalable remote homology classification}},
author = {D. M. Miller and N. Bordin and J. Jeyananthan and V. Waman and M. Heinzinger and C. Orengo},
howpublished = {bioRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.28.747864},
url = {https://www.biorxiv.org/content/10.64898/2026.08.28.747864v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - ContrasTED: contrastive domain embeddings for scalable remote homology classification
AU - D. M. Miller
AU - N. Bordin
AU - J. Jeyananthan
AU - V. Waman
AU - M. Heinzinger
AU - C. Orengo
PY - 2026
DA - 2026/08/28
PB - bioRxiv
DO - 10.64898/2026.08.28.747864
UR - https://www.biorxiv.org/content/10.64898/2026.08.28.747864v1
AB - Protein structure prediction has expanded structural databases to hundreds of millions of domains. Classifying these domains into homologous superfamilies reveals evolutionary and functional relationships that can persist despite low sequence similarity. As the size of structural databases continues to grow, homology classification requires methods that combine scalability with accuracy. Here we present ContrasTED, which uses CATH-supervised center-contrastive learning to project structure-aware embeddings into a domain-level metric space for nearest-centroid superfamily assignment. On a sequence-filtered S20 benchmark (n = 1,028), superfamily assignment accuracy reached 92.9% (1-NN) and 91.4% (nearest centroid), exceeding sequence search, profile HMMs, Foldseek, and a classifier trained on embeddings. The learned latent space separates superfamilies while retaining structural information below 20% sequence identity, with the largest gains among sparsely represented superfamilies. ContrasTED produces 4.67 million new candidate assignments across 3,796 superfamilies in The Encyclopedia of Domains (TED), extending annotation coverage beyond previous structure-based methods.
ER -