search

juanbot / CoExpNets

Co-expression network management based on WGCNA + k-means
Apache License 2.0
16 stars 8 forks source link

Multiple testing correction #10

Closed juanbot closed 5 years ago

juanbot commented 5 years ago

I am responsible for this. We have to define ways for multiple testing. Including two groups of options: level of correction and technique for correction

Level of correction: -No corrections -Correcting within tissue: we correct independently within each network. Within each network, if we have n modules, we consider n tests. -Correcting along all: we correct all p-values. So if we have n networks and m(i) the modules for i-th network, we consider m(1) + m(2) + ... + m(n) tests.

Technique: -Bonferroni -BH -FDR 5%

juanbot commented 5 years ago

Sonia, I just added a new method globalReportOnGenes that will return the category column and two more columns with uncorrected and corrected p-values. This will solve this issue. Could you please try it please?

Please note that the "categories" parameter is of the same length as tissues. For the i-th tissue, its category is categories[i]

globalReportOnGenes(tissues=c("TCTX","WHMT","PUTM","WholeBlood"), categories=c("10UKBEC","10UKBEC","10UKBEC","gtexv6"), genes=c("TREM2","TH","SNCA"))

SoniaRuiz commented 5 years ago

Ok, no problem, I'm working on that.

Regarding the function 'reportOnGenesMultipleTissue()', I have seen that the old 'p-value' field is now called 'fisher'. Is it going to be its definite name? I'm asking that because I need to set the table-header names beforehand.

SoniaRuiz commented 5 years ago

I have updated the testing environment with the new version of CoExp. I'll close this issue once we're sure that this issue is solved.