correlation between protein expression and RNA read counts

Question

[kng]

I have Immunofluorescence(IF)-measured protein expression for each of my samples (6 from two groups - 3 replicates each) and RNA seq counts for each sample.

1) How can I correlate IF value with protein expression for a given gene? Is the correlation coefficient between protein expression and CPM normalized reads for the gene of interest the right way to correlate these data? 2) Can I do the same with all other genes to find out which genes are closely related to protein expression?

Here is what I did to answer Q. 1 and 2

1) IF values = list of protein expression in 6 samples. These are just 6 numeric values co IF_values = c(2.0, 3.99, 44.9, 50.1, 34.2, 1.0, 23.1) (these values are not normalized to anything. These are absolute counts as obtained from IF. RNA_coutns_ABCD = c(334.0, 56.7, 34.4, 33.3, 22.2, 100.5) ie 6 data points from RNA seq counts from each sample (CPM normalized)

then I am computing in R ABCD_coef <- cor(IF Values, RNA_coutns_ABCD)

Can this value be used to infer the correlation between protein expression and RNA expression? The numbers are for example purpose only and are not real.

2) Compute the correlation of all genes (~17K) in the RNA seq count list to the IF values

The genes with the highest correlation values (taking +ve only) are shortlisted as top genes.

Is this the right approach to compare protein expression values to RNA seq?

[Mensur Dlakic]

I don't have an answer to your question.

Out of curiosity, is +ve meant to be positive? If so, how did you come up with that abbreviation? It is not terribly intuitive to me and I had to think for solid 10 seconds to understand from the context what you meant.

[kng]

standard (mathematical) abbreviation: positive (+ve) and negative (-ve)

[Mensur Dlakic]

I have seen all kinds of standard and non-standard abbreviations - and figured out in 2 seconds the other day what YMMV meant - but this is the first time I have seen either +ve or -ve. Good to know.

Answer 1

[Alex Gibbs]

This is a rather difficult question to answer, and perhaps I am not the best person to answer it but here are my thoughts;

Do you have a housekeeping gene/control gene that you used in the IF analyses? If so, perhaps you could normalise your IF gene expression to that gene for each sample and do the same thing with your RNAseq data. This is provided that you have used the same laser settings etc on your microscope across each IF sample.

Sorry that isn't quite an answer you may have been looking for but I hope it helps you get somewhere!

Alex