KS test
- compare how similar distributions are
- less power? sensitive?

t test

independence of values

generate random 'pathways' (use as NULL, instead of what I did) and use for comparison

there have to be tests or metrics that test significance of a ranked list


gradient of the pathway 
- receptor proteins more variable 
- looking at levels (similar to hierarchy?)
-> instead of looking at co-occurrence

Jason Lu
- surmounted 'download KEGG' issue?

KEGG pathways
- XML formats - parse XML files

BioCarta
- downloadable pathways
- Cgap

GSEA - gene-set enrichment analysis