I have a small scRNA dataset (~30 progenitor cells, manually isolated so assumed to be fairly homogeneous) and came across this paper where they compared the expression of previously identified marker genes in their single cells, and then also looked for additional genes that were highly correlated in expression with one of their known markers.
Basically they calculated the correlation between their gene (NF-L) and all other genes, transformed those results using Fisher's Z-transformation, then tested if each correlation was larger than zero while controlling FDR with the Benjamini-Hochberg method to finally produce a list of highly correlated genes.
I was wondering if there were any good R packages to carry out a similar workflow? I managed to perform a 1-to-many correlation (psych::corr.test seemed to be OK?) and was able to carry out the z-transformation on the resulting r-values, however I'm not sure how to go about testing if each correlation is above 0 while controlling FDR.
I'm assuming that what I am attempting to replicate is indeed different to simply using correlations<-corr.test(GOI,t(expression),method = "pearson",adjust = "BH",ci=F) and then taking the correlations that are above 0 and have adjusted p values under the cutoff.
It seems most packages for single cell assume you have a large number of cells, which you would cluster and look for highly variable genes and then look for correlations between, but this does not seem applicable to my dataset in that it is a small, uniform group of cells so the variation is low.
Thanks in advance,