Aug 2026· Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2· pp. 10465-10474· 0 citations· 26 references
TL;DR
CAHS (Context-Aware Homology Search), a lightweight query-time adapter for pHMM search that incorporates learned and biologically motivated signals without changing HMMER's downstream search pipeline or its calibrated E-value reporting, illustrates a practical AI-for-science design pattern.
Abstract
Protein homology search is foundational to bioinformatics: it supports annotation transfer, structure/function inference, and evolutionary analysis over rapidly expanding sequence repositories (e.g., UniProtKB). Profile hidden Markov models (pHMMs), as implemented in HMMER, remain the most widely trusted approach because they provide statistically calibrated E-values; however, their gap behavior is fixed once a profile is trained, despite biological evidence that insertion/deletion tolerance varies across flexible loops and intrinsically disordered regions. We present CAHS (Context-Aware Homology Search), a lightweight query-time adapter for pHMM search that incorporates learned and biologically motivated signals without changing HMMER's downstream search pipeline or its calibrated E-value reporting. Given a query sequence, CAHS computes per-residue representations from a protein language model and a disorder predictor, maps these to profile coordinates, and modulates only match-state transition rows (gap-open and gap-extension probabilities) while preserving Plan7 constraints. We comprehensively evaluate CAHS across six structurally diverse protein families and multi-domain architectures against a 570k-sequence target corpus. CAHS expands detection capability, retrieving thousands of additional remote homologs at relaxed thresholds by maintaining alignment quality through flexible regions. For multi-domain proteins, context-aware modulation resolves 94% of fragmented alignments. Crucially, CAHS preserves hit-set invariance at stringent operating points (E<10-10), demonstrating increased statistical confidence without inflating false positives. Furthermore, sharper statistical distinction between homologs and background noise during early filter stages yields up to a 3.87× acceleration in end-to-end wall-clock time on high-performance computing clusters. Overall, CAHS illustrates a practical AI-for-science design pattern: augmenting a trusted probabilistic model with query-specific learned signals to improve interpretable, reproducible inference in data-rich biology.
A pipeline reformulating kinase-substrate modeling as a Bayesian inference problem is presented and it is revealed that the interaction types and distances to the catalytic pocket significantly influence pathogenicity scores.
Jinyuan Hu, Shimian Li, Yue Xue et al.· Journal of Chemical Informat...· 0 citations
BACKGROUND
Protein sequence and structure similarity-based search is an important task, which underpins protein annotation, evolutionary analysis, large-scale functional inference, and the exploration of the protein "dark space". The rapid growth of sequence and predicted structure databases has spurred diverse search methods, yet their evaluation remains limited to fold-level similarity and inconsistent benchmarking protocols.
RESULTS
We present a comprehensive benchmark for protein sequence and structure search. Using this framework, we evaluate 14 representative methods spanning sequence alignment, structure alignment, and representation-based approaches across multiple biologically relevant scenarios. Our results show pronounced and context-dependent differences among methods. Structure alignment methods excel at detecting fold-level and geometric similarity, while representation-based searching approaches show advantages in capturing functional similarity under low sequence identity and robustness to predicted structures. Notably, all evaluated methods show limited effectiveness on intrinsically disordered proteins.
CONCLUSIONS
This benchmark establishes a standardized framework for evaluating protein similarity search methods, providing a practical resource for method selection and a foundation for the development of next-generation approaches capable of addressing diverse homology search challenges.
Yuan Liu, Yingquan Zhou, Yan Huang et al.· Genome Biology· 1 citation
Identifying homologous proteins across deep evolutionary distances remains a major challenge because sequence and structural similarity progressively become undetectable over time. Although protein-protein interactions (PPIs) are often constrained by function and evolution, whether conserved interaction interfaces can provide an independent signal for homology detection has remained largely unexplored owing to the computational cost of proteome-scale interaction prediction. Here we introduce HInt (Homology by Interaction), an accelerated AlphaFold-based framework that enables practical proteome-scale PPI prediction through biologically informed pre-filtering and optimised high-throughput structure modelling. Using HInt, we establish interaction-based similarity as a third axis of homology detection. We show that conserved interaction interfaces reveal homologous relationships that remain inaccessible to conventional sequence- and structure-based approaches. Application of HInt to both prokaryotic and eukaryotic systems, together with experimental validation, uncovered a previously unrecognised VirB5 pilus-tip protein in the F-plasmid type IV secretion system and a previously unannotated F-box-like protein in the Saccharomyces cerevisiae ubiquitin-proteasome system. By enabling practical proteome-scale interaction screening, HInt provides a general framework for uncovering hidden homologues and expands the conceptual landscape of protein homology inference.
Quentin Rouger, P. Paillard, Manon Thomet et al.· bioRxiv· 0 citations
Accurate multiple sequence alignment (MSA) is central to understanding protein evolution, structure, and function. We present BioMatics 1.0, a novel MSA algorithm that applies optimal transport principles through the Wasserstein first-order distance function to align amino acid distributions across positions, enabling refined detection of structural and evolutionary patterns. Unlike conventional score-based methods, BioMatics 1.0 constructs profile-to-profile alignments using Earth Mover's Distance over per-position frequency vectors, guided by BLOSUM62 log-odds similarity. This is complemented by entropy-adaptive gap penalties that dynamically modulate alignment behavior in variable or weakly conserved regions. Benchmark evaluations across curated datasets spanning conserved domains, structural motifs, and heterogeneous families demonstrate that BioMatics 1.0 outperforms widely used tools in column score (CS) accuracy and achieves competitive or comparable sum-of-pairs score (SPS) results. Its architecture prioritizes residue-level alignment precision, yielding results that are particularly informative for downstream tasks such as phylogenetic reconstruction and structure-informed modeling.
Orkid Coskuner-Weber, Yusuf Emre Ari, Yildiray Efe Berberoglu et al.· Proteins: Structure, Functio...· 0 citations
The functional annotation of genes in non-model organisms remains a significant challenge in computational biology, with 20-70% of sequenced genes lacking characterized functions. Traditional homology-based methods are often costly and strongly dependent on high sequence similarity. This study presents Homo-RAG, a framework for large language model-based gene function prediction that integrates homology-guided multi-hop retrieval with evidence-aware ranking. The framework exploits biological relationships between zebrafish and human orthologs to guide evidence acquisition from ZFIN, UniProt, and PubMed through hybrid dense and lexical retrieval. An Evidence Confidence Score (ECS) integrates semantic relevance, entity matching, orthology information, source reliability, and literature association signals to refine the ranking of retrieved evidence. Extensive evaluation across 150 queries and 7,200 retrieved documents shows that evidence weighting parameter of lambda=0.50 improves NDCG@10 to 0.9879 and MRR to 0.99, while retrieving relevant evidence for 99.33% of queries. Furthermore, 80% of the retrieved documents are query-exclusive, indicating that evidence quality complements rather than replaces retrieval relevance. These findings establish Homo-RAG as a practical and robust framework for reliable, evidence-grounded gene function prediction in understudied organisms. The study addresses important limitations of conventional annotation pipelines while identifying opportunities for future improvements in evidence features and attribution mechanisms.
WASP highlights how structural homology can systematically discover annotations missed by sequence-based approaches, predicting protein functions from AlphaFold structures using network-based structural homology and filling metabolic model gaps by mapping 75-100% of orphan reactions.
