Virtual cells: progress and challenges from mechanistic modeling to data-driven integration

Review (Ma et al.; senior authors Jiang Q and Chu Y; Briefings in Bioinformatics 2026) that uses state encoding, state transition and observation decoding as a common abstraction across mechanistic models, data-driven foundation and perturbation models, multimodal systems and spatial/multiscale simulation, and examines model boundaries, data foundations, resource disclosure, out-of-distribution evaluation, causal evidence and experimental validation. The authors note that independent benchmarks show parameter count and training-set size do not automatically yield robust out-of-distribution generalization.

Authors / organisation

Ma N, Wang Z, Yang X, Yang G, Yang R, Wang Z, Jiang Q, Chu Y.

Venue

Briefings in Bioinformatics

Date

2026-10-18

Type

paper

Year

2026

DOI

10.1093/bib/bbag562

PMID

42893561

Verification

Crossref (journal-article, vol 27, article bbag562, published online 2026-10-18, issue date 2026-09, CC BY 4.0, 8 authors, abstract) + PubMed PMID 42893561 fetched 2026-10-20; full text not read

ReviewAI virtual cellMechanistic & agent-based modelFoundation modelBenchmark & validation

Open in the searchable library → · JSON