Load data because I'm a noodle and accidentally cleared data from my previous workspace!
load("/home/vanessa/Desktop/outputup.Rda")
load("/home/vanessa/Desktop/outputdown.Rda")
load("/home/vanessa/Desktop/number_ME.Rda")
thresh = 0.05
I (fingers crossed) hope that these distributions are more realistic!
# Like the fairies say in sleeping beauty - make it pink! Make it blue!
hist(OUTPUTup@adjp, main = "Adjusted shapley p values for UP set", col = "pink")
hist(OUTPUTdown@adjp, main = "Adjusted shapley p values for DOWN set", col = "blue")
Promising… but specifically - are there overlap in probes for the two sets? There technically shouldn't be…
probesup = colnames(probes)[which(OUTPUTup@adjp < thresh)]
probesdown = colnames(probes)[which(OUTPUTdown@adjp < thresh)]
any(probesup %in% probesdown)
## [1] FALSE
Shweet! Could this be the correct way to do it? What are the size of the two sets…
length(probesup)
## [1] 1232
length(probesdown)
## [1] 152
Very different! Ok, I had some other ideas for ways to get subsets, but I think this might work! I'm going to stop and send this to you! If you think this is the right way to go, I'll update my scripts, re-generate the sets, and then use them for GSEA. I've already run all of Leticia's GSEA analyses across the 8 nextbio expression sets, and I'd like to create a database with my gene probes and run the same expression sets against a “brainterm” database and then write a script to compile the results. I will want to talk about with you how to adjust for multiple comparisons across allll those analyses! I asked Maude and she said she could look into it, but I haven't asked again yet :) I really like how we can all work together on this! That means that we have a good lab.