Profile count wrong in scenario where a WES study is paired with a gene panel study

[alisman]

http://localhost:8082/study/summary?id=msk_impact_2017%2Clihc_tcga_pan_can_atlas_2018

This is two studies, one WES (pan-can) and the other has a gene panel (IMPACT). Take a gene that is NOT in the gene panel. Samples in the impact study are NOT profiled for this gene. Therefor the number of profiled samples in this screenshot is wrong, as is the denominator of the frequency calculation. The correct number of profiled samples is the total number of samples in the WES study (372). FYI, until Rob completes his work reimporting the studies, there is a further complication with the calculation of that number, but we ignore that for now. It will soon be fixed.