Categorization and analysis of 16 computational methods for estimating cell potency from single-cell RNA-seq data

Inferring hidden cellular dynamics from static single-cell RNA sequencing (scRNA-seq) data remains a critical challenge. Beyond pseudotime trajectories and RNA velocities, cell potency provides another perspective by estimating cellular differentiation potential from gene expression levels. We review 16 computational methods that define cell potency under different, sometimes contrasting, assumptions. On the basis of how they summarize gene expression into a potency measure, we categorize them into average-based, entropy-based, and knowledge-based methods. Using a unified set of mathematical notations, we detail their methodologies, explain key similarities and differences within and between categories, and summarize usage complexities. Given the diversity of methods, community benchmark datasets at single-cell resolution and careful sensitivity analyses are crucial for evaluating methods from multiple angles and potentially improving stability and generalizability. This article provides a basis for method benchmarking and highlights the remaining challenge of comparing methods for inferring pseudotime trajectories, RNA velocities, and cell potency.