OscoNet

OscoNet infers oscillatory gene networks from snapshot single-cell RNA sequencing (scRNA-seq) data to detect genes with periodic (sinusoidal) expression and recover cluster-specific pseudo-time ordering.

Key Features:

• Improved Optimization Scheme: Enhances the optimization framework originally proposed by Oscope to more accurately detect co-oscillatory gene pairs.
• Statistical Rigor: Incorporates a well-calibrated non-parametric hypothesis test to select oscillatory genes at a specified false discovery rate (FDR).
• Enhanced Sensitivity: Demonstrates superior sensitivity compared to Oscope, identifying larger sets of known oscillators while avoiding ambiguous thresholds.
• Pseudo-Time Estimation: Implements an advanced pseudo-time estimation technique that accurately recovers cell order for each gene cluster and is computationally faster than the extended nearest insertion method.

Scientific Applications:

• Periodic process analysis: Study of periodic gene expression processes such as the circadian clock and cell cycle in single-cell data.
• Oscillatory network inference in scRNA-seq: Identification and characterization of gene modules exhibiting sinusoidal oscillations from snapshot scRNA-seq.

Methodology:

Analyzes snapshot scRNA-seq data to identify genes with sinusoidal oscillations using an improved Oscope-based optimization framework, applies a non-parametric hypothesis test controlling FDR to select oscillatory genes, infers co-oscillatory gene pairs/networks, and estimates pseudo-time per gene cluster with a method faster than the extended nearest insertion approach.