scEpath: energy landscape-based inference of transition probabilities and cellular trajectories from single-cell...

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Single-cell RNA-sequencing (scRNA-seq) provides high-resolution data for studying cellular processes.
• Inferring cell state transition paths and probabilities from scRNA-seq data is crucial but challenging.

Purpose of the Study:

• To present scEpath, a novel algorithm for reconstructing developmental trajectories from scRNA-seq data.
• To enable robust inference of cell state transition probabilities and lineage relationships.
• To identify key genes and regulatory networks involved in cell fate decisions.

Main Methods:

• scEpath calculates energy landscapes using 'single-cell energy' and distance-based measures.
• It constructs probabilistic directed graphs to model cell state transitions.
• The approach is largely unsupervised and robust to variations in gene set size.

Main Results:

• scEpath accurately reconstructs pseudotemporal orderings and cell lineage relationships.
• The algorithm identifies marker genes and gene expression patterns associated with cell state transitions.
• Applications revealed a cell-cell communication network in early human development and identified novel transcription factors for myoblast differentiation.

Conclusions:

• scEpath provides a robust and high-resolution method for analyzing cell state transitions and developmental trajectories.
• The algorithm offers new insights into cell fate determination and regulatory mechanisms during differentiation.
• scEpath facilitates the identification of temporal dynamics and transcriptional programs along branched lineages.