scTrends: automated classification and strength quantification of gene expression trends along pseudotime in single-cell RNA-seq

BackgroundPseudotime inference has become a standard approach for reconstructing dynamic biological processes from single-cell transcriptomic data. However, after a pseudotemporal ordering has been established, systematically identifying and interpreting gene expression trends along pseudotime remains challenging. Existing approaches often rely on clustering-based heuristics or subjective parameter choices, which can compromise interpretability, reproducibility, and scalability in large datasets. ResultsWe present scTrends, an automated and interpretable framework for gene-level trend classification and strength quantification along a given pseudotime trajectory. Importantly, scTrends does not perform pseudotime inference; instead, it operates downstream of established pseudotime methods to characterize expression dynamics once a temporal ordering is available. scTrends models pseudotime-binned gene expression profiles using generalized additive models and assigns genes to predefined temporal trend categories through a hierarchical, rule-based procedure combined with empirical significance testing, data-adaptive parameter selection, and quantitative assessment of trend strength. This enables simultaneous identification of the direction, shape, and magnitude of gene expression changes along pseudotime. We applied scTrends tothree distinct datasets: human PBMC, human brain, and mouse pancreas, using three different pseudotime inference methods (CytoTRACE v2, Monocle3, and scVelo, respectively). scTrends systematically characterized gene expression dynamics during T cell differentiation, oligodendrocyte precursor differentiation, and pancreatic endocrine cell maturation. The analysis revealed diverse monotonic, non-monotonic, and complex expression patterns, with varying strengths, consistent with known biological processes. Benchmarking analyses further demonstrate that scTrends is computationally efficient and scalable to large single-cell datasets, with modest memory requirements, making it suitable for diverse applications across a range of biological systems. ConclusionsscTrends provides a systematic, automated, and resource-efficient solution for gene-level trend analysis in single-cell pseudotime studies, enabling reproducible characterization of dynamic expression patterns across diverse biological systems.