ArchVelo: Archetypal Velocity Modeling for Single-cell Multi-omic Trajectories

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Inferring dynamic cellular processes from static single-cell measurements is a significant challenge.
• Existing methods struggle with accurate gene regulation and cell trajectory inference.

Purpose of the Study:

• Introduce ArchVelo, a novel computational method for modeling gene regulation and cell trajectories.
• Leverage single-cell simultaneous chromatin accessibility (scATAC-seq) and transcriptomic (scRNA-seq) profiling for enhanced inference.

Main Methods:

• ArchVelo models chromatin accessibility as archetypes (shared regulatory programs).
• It dynamically models the influence of these archetypes on transcription.
• The method integrates scATAC-seq and scRNA-seq data.

Main Results:

• ArchVelo demonstrates improved inference accuracy and latent time alignment compared to previous methods.
• It successfully identifies underlying transcription factor activity.
• Applied to developing mouse brain, human hematopoiesis, and CD8 T cell responses, it revealed distinct differentiation and proliferation trajectories.

Conclusions:

• ArchVelo provides a principled framework for analyzing dynamic gene regulation in multi-omic single-cell data.
• The method enables deeper insights into cellular processes across diverse biological systems.
• It identified a novel differentiation trajectory in CD8 T cell progenitors relevant to immune responses and immunotherapy.