correlation between genes

Question

[1769mkc]

I have different cell type total of 5 ,im doing a pairwise comparison, I would like to show the correlation in pairwise condition between the genes. I want to plot how these certain genes are expressed. So would be be suffice to show simple correlation based calculation or i have to do something else.

Any suggestion would be appreciated.

[Santosh Anand]

Also have a look at the corrplot package https://cran.r-project.org/web/packages/corrplot/vignettes/corrplot-intro.html

[1769mkc]

I use corrplot often but my question was would it be enough to make a correlation to show which genes are coexpreseed or i have to go for WGCNA kind of thing ?

[h.mon]

Your question could be better phrased. For finding co-expressed genes, I think you should go for PCIT, WGCNA or some similar method.

[1769mkc]

sorry for my confusing words but thats what i wanted to ask

Answer 1

[Devon Ryan]

It sounds like a simple pairwise correlation with cor() would suffice.

[h.mon]

Probably better to also use method="spearman" with cor(), just to be safe and avoid spurious correlations.

[Gema Sanz]

Hi, following this discussion, I've been trying to use cor() after transposing my gene/samples matrix from microarray experiment (so genes are as columns and samples are as rows) but I'm not sure how to define "y", I get this error:

cor<- cor(tm, use="pairwise.complete.obs","spearman")  Error in
 cor(tm, use = "pairwise.complete.obs", "spearman") :    'y' must be
 numeric

[Devon Ryan]

Make tm a matrix.

[Gema Sanz]

it is a matrix already, I did

tm <- as.matrix(tm)

but the only way to avoid the error is if I set "y<-tm", then I get a correlation table of 900x900 with the correlation of each gene with every other gene but I'm not sure what I'm comparing because y should be 6 samples... I don't know

anyhow I just posted the whole question here

How to calculate moderated z-score (MODZ)

[Devon Ryan]

Oh, you need method="spearman", not just "spearman". R is seeing "spearman" as the thing you want to correlate tm with otherwise.