eXplainable PRS (XPRS)

Step 1: Preprocessing

  • Align risk alleles in the genotype.file and PRS.scoring.files, with adjustment of beta sign.
  • Conduct Gene mapping, which consists of three stages as follows.
    1. Positional mapping: based on genomic position.
    2. Combine Snp-to-Gene (cS2G) mapping: based on various linking techniques such as expression quantitative trait loci (eQTL), enhancer-gene linking, and promoter capture Hi-C (PCHI-C).
    3. GWAS p-value and Snp heritability-based mapping: regional mapping based on p-value in user-provided GWAS summary statistics or SNP heritability for unmapped SNPs, within a default window size of 500kb.

Step 2: Calculation Contribution Score and Attributed Value

  1. Gene Contribution Score (CSgene)
    • This score quantifies how much each gene contributes to increasing the individual PRS. It is calculated by summing the weighted risk alleles mapped to genes.
  2. The Attributed Value of gene (Agene)
    • This value identifies genes that elevate disease risk by comparing the gene's contribution score to the average contribution score in population, using SHAP.
  3. SNP Contribution Score (CSSNP)
    • This score quantifies the contribution of each SNP to the individual's PRS.
    • x in the equation is a number of risk SNPs (x, where x = 0,1,2).

Step 3: Visualization

  • The computed values from step 2 are presented at population and individual levels for enhanced comprehension.
  • The risk factors were indicated using the Manhattan plot for population and individual, LocusZoom-like plot for individual.