Cancer cell dormancy is a clinically consequential yet computationally under-defined phenomenon characterised by reversible growth arrest and delayed disease recurrence. Although advances in single-cell and multi-omic profiling have improved detection of dormant and persister populations, their molecular identity and dynamical behaviour remain difficult to resolve. In this review, we examine how computational methods have been applied to infer dormant cell identity, heterogeneity, microenvironmental regulation, state transitions, and reactivation dynamics. We highlight how single-cell transcriptomics, lineage tracing, spatial profiling, and integrative multi-omic analyses reveal substantial context-dependent variability, undermining the notion of a universal dormancy signature. We further discuss emerging mathematical and statistical frameworks to model the awakening from dormancy, alongside approaches linking dormancy-associated features to clinical outcomes. Recurring challenges include fragmented operational definitions, rare-state detection, cross-study incompatibility, and the use of snapshot data to interrogate inherently temporal processes. We argue that progress will depend on computational frameworks that treat dormancy as a dynamic, multi-scale systems problem rather than a static cell-type classification task.
Spink et al. (2026) studied this question.