Make sample select more hierarchical

[keller-mark]

After loading PanCanAtlas data, probably should modify sample select to make it more hierarchical (e.g. by project or by cancer type) since there will be ~80 sample cohorts