In silico approaches for discovering microbial antiviral defense systems.

Abstract

Prokaryotes possess a remarkably diverse and dynamic repertoire of antiviral defense systems, enabling them to withstand phage predation. However, their frequent horizontal gene transfer, extensive sequence diversity, modular genomic organization, and rapid evolution make purely experimental discovery challenging. Coupled with the massive influx of microbial genomes from high-throughput sequencing, computational strategies have become indispensable complementary tools that can enhance the efficiency and scope of defense systems discovery. In this review, we categorize computational approaches into four major strategies: (i) Sequence homology-based methods, which reliably annotate known defense systems through protein sequence similarity but are limited in detecting highly divergent or novel systems; (ii) Structure-guided approaches, which leverage conserved protein folds to uncover remote homologs and single-gene defense proteins, providing sensitivity beyond sequence-based identification, albeit at high computational cost; (iii) Genomic context-based strategies, which exploit gene co-localization and defense islands to uncover multi-gene defense clusters and previously uncharacterized defense modules; and (iv) Artificial intelligence-powered methods, which integrate sequence-derived embeddings with genomic context information to predict low-homology proteins and reconstruct candidate defense systems at scale, enabling discovery of novel systems beyond the reach of conventional approaches. We further discuss emerging tools and frameworks, such as the conserved gene cluster discovery tool and genomic foundation models, which hold strong potential to extend conventional approaches for identifying novel defense systems and supporting the generative design of synthetic modules. By comparing methodological principles, strengths, and limitations, this review provides a practical framework for the systematic exploration of microbial immune systems, guiding applications such as rational phage therapy, microbiome engineering, and synthetic biology.