Quantifying the clusterness and trajectoriness of single-cell RNA-seq data

Area of Science:

• Computational Biology
• Single-Cell Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) analysis commonly employs clustering and trajectory inference methods.
• These methods can yield divergent data interpretations, complicating biological insights.
• A need exists to objectively assess data structure for appropriate analysis selection.

Purpose of the Study:

• To introduce quantitative scores for 'clusterness' and 'trajectoriness' in scRNA-seq data.
• To provide objective metrics for distinguishing between cluster-like and trajectory-like data structures.
• To guide the selection of appropriate computational analysis for scRNA-seq data interpretation.

Main Methods:

• Development of multiple quantitative scores based on pairwise distance distribution.
• Application of persistent homology, vector magnitude, Ripley's K, and connectivity degrees.
• Validation using simulated and real-world scRNA-seq datasets.

Main Results:

• Proposed scores effectively differentiate between cluster-like and trajectory-like data patterns in simulations.
• Scores demonstrate utility in guiding analysis choices for real scRNA-seq datasets.
• Quantification provides objective criteria for selecting clustering versus trajectory inference.

Conclusions:

• The developed 'clusterness' and 'trajectoriness' scores offer a robust method for data structure assessment.
• These scores aid researchers in choosing the most biologically relevant analysis for scRNA-seq data.
• Objective quantification improves the reliability and interpretability of single-cell data analysis.