mRNA expression levels association

Question

[sarahmanderni]

Hi all,

I was wondering what is the most efficient way of investigating the association of expression levels of two different genes in an RNA-seq experiment. I mean simply, if you want to check if gene A and gene B are associated to each other for different samples. Do you think a scatter plot could be meaningful here?

Thanks!

[Na Sed]

You can calculate 'correlation' between them to see how they are correlated.

Answer 1

[TriS]

in R you could do (not tested):

lm_genes <- lm(geneB ~ geneA)
pv <- summary(lm_genes)$coefficients[2,"Pr(>|t|)"]
r2 <- summary(lm_genes)$r.squared
coeff <- cor(geneA, geneB)
plot(geneA, geneB)
abline(lm_genes, col="red")
legend("top",legend=paste(paste("pv =",pv),paste("R^2 =",r2),paste("Coeff =",coeff),sep="\n")

[sarahmanderni]

Thanks. I also was thinking of calculating the correlation and showing it with a scatter plot. But I am wondering does this have any biological meaning? I mean suppose I got high correlation value does this say anything to the biologists?

[TriS]

the meaning of what you do is connected to the question that you are asking.

if you take prostate samples and you measure the androgen receptor (AR) and the prostate specific antigen (PSA) expression you will see that they positively correlate...why? because AR targets PSA and promotes its transcription. this is biologically meaningful.

now, gene X correlates with gene Y...ok...is that known? is this biologically meaningful...well, do papers show that gene X targets Y? if not it'll be up to you to show through knockdown experiment, pharmacological inhibition etc etc...this means that your RNASeq results can be used to generate hypothesis that can be tested in vitro. however...it seems to me that you need to sit down, look at your results and maybe discuss with your PI what the next step can be. this can be validating your RNASeq results and/or developing new hypothesis based on what you find