SwitchTFI: identifying transcription factors driving cell differentiation.

UNLABELLED: Many methods exist that infer cell differentiation trajectories from single-cell RNA sequencing data, but only few determine which mechanisms drive the inferred differentiation dynamics. To close this gap, we developed the algorithm and Python package SwitchTFI. Utilizing regression stump learning, permutation-based family-wise error rate control, and node centrality measures, SwitchTFI identifies differentiation-driving gene regulatory networks and the key transcription factors involved in them. Comprehensive tests on pancreatic endocrinogenesis, erythrocyte differentiation, and T cell exhaustion datasets show that SwitchTFI can rediscover known differentiation factors, that it can discover novel biologically plausible hypotheses, and that it compares favorably to competitor methods. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13059-025-03876-0.      «

UNLABELLED: Many methods exist that infer cell differentiation trajectories from single-cell RNA sequencing data, but only few determine which mechanisms drive the inferred differentiation dynamics. To close this gap, we developed the algorithm and Python package SwitchTFI. Utilizing regression stump learning, permutation-based family-wise error rate control, and node centrality measures, SwitchTFI identifies differentiation-driving gene regulatory networks and the key transcription factors invo...     »