- Use different aggregation statistics per pathway
- E.g. a gene-level: Proportion of genes altered (in case one gene w/ multiple hits dominates)
- E.g. use a hypergeometric P-value to illustrate significance.
- E.g. use proportion of disruptions out of total possible
Feedback
- Add gene expression aggregation - how many outlier genes in pathway?
- Add in your own geneset
- Use predicted functional impact of a mutation instead of just presence/absence?