scTenifoldKnk

scTenifoldKnk performs virtual gene knockouts on single-cell gene regulatory networks reconstructed from wild-type (WT) single-cell RNA sequencing (scRNAseq) data to predict gene function and regulatory perturbations.

Key Features:

• Virtual Knockout Capability: Simulates deletion of a target gene by setting the outdegree edges of the target gene in the adjacency matrix to zero.
• Gene Regulatory Network Construction: Constructs single-cell gene regulatory networks (scGRNs) from WT scRNAseq datasets.
• Identification of Perturbed Genes: Compares the modified scGRN with the original WT scGRN to identify differentially regulated genes termed virtual-knockout perturbed genes.
• Manifold Alignment Technique: Employs manifold alignment to align the reduced (post-knockout) GRN with the original network to enhance detection of differential regulation.
• Predictive Power: Shows strong correlation between its virtual knockout predictions and findings from real-animal KO experiments.
• Resource Efficiency: Relies solely on WT scRNAseq data, eliminating the need for additional KO sample data.

Scientific Applications:

• Gene Function Elucidation: Predicts and explores gene roles and regulatory effects at single-cell resolution.
• Prioritization of Experimental Targets: Ranks candidate genes for follow-up real-animal KO experiments based on predicted perturbations.
• Predictive Modeling: Enables in silico simulation of genetic interventions to anticipate experimental outcomes.

Methodology:

Input WT single-cell RNA sequencing (scRNAseq) data; construct a gene regulatory network (scGRN) from the WT data; execute a virtual knockout by modifying the adjacency matrix (setting the target gene's outdegree edges to zero); apply manifold alignment to compare the modified (reduced) GRN with the original WT GRN and identify differentially regulated genes (virtual-knockout perturbed genes) for functional inference.