XGSEA

XGSEA predicts enrichment significance of gene sets in a target species (e.g., human) from gene expression data in a source species (e.g., mouse) to enable cross-species gene set enrichment analysis when direct measurements in the target species are not feasible.

Key Features:

• Cross-species analysis: Predicts enrichment significance for target-species gene sets using source-species gene expression when direct measurement in the target species is infeasible.
• GSEA on source species: Performs Gene Set Enrichment Analysis on the source species to obtain enrichment scores and p-values for gene sets.
• Domain adaptation: Uses domain adaptation to represent and bridge relationships between source and target species gene sets.
• Regression prediction: Applies regression models to predict p-values of target-species gene sets from the adapted representations.
• Validation and performance: Validated on four real datasets and reported to significantly outperform three baseline methods.
• Case study — ATAC-Seq and T cell pathways: Demonstrated identification of human pathways related to T cell dysfunction and reprogramming using mouse ATAC-Seq data.

Scientific Applications:

• Translational research: Infers human pathway enrichment and gene-set relevance from model organism data when human experiments are constrained by ethics or safety.
• Pathway and target identification: Supports identification of biological pathways and potential therapeutic targets across species.
• Cross-species inference: Enables application of model-organism gene expression results (e.g., mouse) to human biological questions, including studies of complex diseases.

Methodology:

Performs GSEA on the source species to obtain enrichment scores and p-values; applies domain adaptation to relate source and target gene-set representations; and fits regression models to predict target-species gene-set p-values from the adapted representations.