Giters

UrszulaCzerwinska / DeconICA

Deconvolution of transcriptome through Immune Component Analysis

Home Page:https://urszulaczerwinska.github.io/DeconICA/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

t.test function

UrszulaCzerwinska opened this issue · comments

commented

implement t.test on A matrix selected samples

commented

Experimental implementation runs the distribution comparison tests i.e. t-test, kruskal-wallis any other classic test + edgeR::exactTestto compute p-value between given threshold of distribution given by A matrix of ICA for all genes

so far the t-test results for TCGA BRCA looks like this:
t-test

the density distribution:
density_t_test

comment : I don't find it very interesting, maybe we should consider log or cumulative distribution

commented

After meeting with AZ:
cut the plot for p-val < 0.1
Perform it on log for t.test - no problem
Perform it for all ICs and select the ones with lower values as in the first plot here -> there will be confounding factors

commented

For 2 ICs of Metabric the function takes 19 min, which means that for the whole metabric (24000 genes) it will take more than 15h

commented

do.call slowers down the function a lot. I found better one match.funthat can be applied for the same purpose.

And I use t.test as a special case (then it is faster).
For metabric it should take 28 min