GSPA

GSPA maps gene sets into a low-dimensional embedding that reflects protein-protein interaction (PPI) network topology to improve identification and prioritization of functionally and disease-associated gene sets from gene expression datasets.

Key Features:

• Latent Embedding Space: Represents genes and gene sets in a low-dimensional embedding that captures PPI network topology and inter-gene dependencies.
• Improved Analytic Power: Incorporates PPI network topology to enhance detection of disease-associated pathways and improve reproducibility of enrichment statistics for similar gene sets.
• Statistical Simplicity: Can be reduced to a version of traditional gene set enrichment analysis via a single user-defined parameter.

Scientific Applications:

• Disease Pathway Identification: Applied to disease-matched gene expression datasets to identify disease-associated pathways.
• Drug Association Discovery: Used to identify novel drug associations with SARS-CoV-2 viral entry.
• Clinical Validation: Predictions were retrospectively validated using claims data from 8 million patients, identifying gabapentin as a risk factor and metformin as a protective factor for severe COVID-19.

Methodology:

Integrates gene set collections and protein-protein interaction (PPI) networks into low-dimensional gene embeddings and extends traditional gene set enrichment analysis into this latent embedding space; the method can be reduced to a GSEA-like analysis via a single user-defined parameter.