SVP

SVP predicts cell states and their variability in single-cell or spatial omics data by computing activity scores of gene sets and identifying spatially variable features using multiple correspondence analysis and random walk with restart algorithms.

Key Features:

• Gene Set Activity Scoring: Calculates activity scores of gene sets including cell marker genes, KEGG pathway, GO ontology, gene modules, transcription factor or miRNA target sets, and Reactome pathway using random walk with restart on a nearest neighbor graph built in MCA space, with scores weighted by hypergeometric test results from the original expression matrix.
• Spatial Variable Feature Detection: Provides global and local spatial autocorrelation methods to identify spatially variable gene sets or other features in single-cell or spatial data.
• Spatial Colocalization Analysis: Detects spatial colocalization between features using spatial autocorrelation approaches.
• Interoperability: Built on the SingleCellExperiment class for compatibility with existing computing ecosystems.

Scientific Applications:

• Cell State Prediction: Predicts cell states and their variability from single-cell or spatial omics data through gene set activity scoring.
• Pathway and Gene Set Analysis: Evaluates the activity of biological pathways and gene sets including KEGG, GO, Reactome, transcription factor targets, and miRNA targets in single-cell and spatial contexts.
• Spatial Transcriptomics Analysis: Identifies spatially variable features and spatial colocalization patterns in spatial omics datasets.

Methodology:

SVP builds nearest neighbor graphs using distances between cells and cells, features and features, and cells and features in the space of multiple correspondence analysis (MCA), applies random walk with restart algorithm to calculate gene set activity scores, weights these scores using hypergeometric test results from the original expression matrix, and employs global and local spatial autocorrelation methods to identify spatially variable features.