scKINETICS: inference of regulatory velocity with single-cell transcriptomics data.

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Transcriptional dynamics are crucial for biological systems, from development to disease.
• Current RNA velocity methods do not capture regulatory drivers of gene expression changes.
• Understanding gene regulatory networks is key to deciphering cellular processes.

Purpose of the Study:

• To introduce scKINETICS, a novel dynamical model for gene expression analysis.
• To simultaneously infer transcriptional velocities and gene regulatory networks.
• To provide mechanistic insights into transcriptional dynamics.

Main Methods:

• scKINETICS employs an expectation-maximization approach.
• It integrates epigenetic data, gene coexpression, and phenotypic manifold constraints.
• The model learns the impact of regulators on target genes.

Main Results:

• scKINETICS successfully models gene expression dynamics in an acute pancreatitis dataset.
• It recapitulates known acinar-to-ductal transdifferentiation pathways.
• Novel regulators of pancreatic processes, including tumorigenesis factors, were identified.

Conclusions:

• scKINETICS enhances existing velocity methods for more interpretable models.
• The approach provides mechanistic understanding of gene regulatory dynamics.
• The method has broad applicability in systems biology and disease research.