The architecture of computational antiviralism: a multi-scale framework from molecular targeting to viral ecosystem engineering.

The recurrent outbreak of viral pathogens and the possibility of the new pandemics demand the transition to the predictive and integrative computational frameworks instead of reactive one. This review describes computational antiviralism as an integrated approach that uses artificial intelligence (AI), virtual screening, and molecular design tools to identify antiviral targets at the molecular level. We compare deep learning-based structure models with physics-based molecular dynamics (MD) with network pharmacology to describe virus-host interactome dynamics. We also evaluate systems virology strategies that combine the transcriptomic, proteomic, and metabolomic data in order to solve infection-induced cellular reprogramming. The framework is not confined to the molecular, but includes evolutionary phylogenomics, epidemiological modelling of the zoonotic spillover, and climate-guided forecasting of cross-species transmission of viruses. We consider such key issues as assay heterogeneity, interpretability of models, and management of autonomous laboratory systems. Importantly, we explicitly acknowledge that no true end-to-end validated multi-scale antiviral pipeline currently exists; the framework is presented as a forward-looking research agenda with clearly defined open challenges. Collectively, this synthesis will bring computational antiviralism as an anticipatory field of study that can catalyze the broad-spectrum antiviral discovery, as well as providing preemptive countermeasures to emergent viral challenges through coordinated molecular, cellular and ecosystem-level interventions.