Run BayesPrism

[lahuuki]

• [x] Install on JHPCE
• [x] Run Example
• [x] Run on DLPFC dataset - FULL
• [x] Run on DLPFC dataset - MeanRatio top25
• [x] Run on DLPFC dataset - 1vALL top25

[lahuuki]

https://github.com/Danko-Lab/BayesPrism R pacakge with (unsecure) webtool

[lahuuki]

Notes from BayesPrism Docs

• "We recommend the use of unnormalized and untransformed count data"
• "To reduce batch effects and speed up computation, we perform deconvolution on protein coding genes"
• filter out genes expressed in <5 cells, Rb, Mrp, other_Rb, chrM, MALAT1, chrX, chrY
• assign.category() seems useful for annotating genes
• Marker selection with pairwise t-test get.exp.stat() & select.marker()
• "we would like to have sufficient number of genes selected for each cell type (>50)"