Hi everyone! Apologies in advance for the long winded question!
Basically, I have been accessing publicly available data in order to shed some light on the relationship between the expression of CD200 (which is a cell surface protein that we are interested in) and cancer prognosis. The other question I also wanted to ask was whether or not there was a correlation between CD200 expression and particular immune cells, such as NK cells within the cancers available.
To do this I accessed two databases: 1) PRECOG and 2) CIBERSORT. PREdiction of Clinical Outcomes from Genomic profiles (PRECOG, http://precog.stanford.edu), has 30,000 expression profiles from 166 cancer expression data sets covering 39 distinct malignancies; and CIBERSORT (Cell type Identification By Estimating Relative Subsets Of known RNA Transcripts) can be used to characterise cell composition of complex tissues from their gene expression profiles.
In the PRECOG paper they ran all of their PRECOG expression profiles through CIBERSORT to produce information on immune cell profiles, etc, which I've got access to.
Basically, l used the PRECOG database to extract z scores for CD200 expression in all of the tumours available. The z-score shows whether or not CD200 expression was associated with good or bad prognosis. A negative z score indicates good prognosis whereas a positive z score indicates bad prognosis. I then extracted the immune profile data that they generated in the PRECOG paper (iPRECOG excel file in z scores, attached) which again contains z scores showing good or bad prognosis in a variety of different cancers for activation or inactivation of particular immune cells. What I've done is combine the two z score sets to see if there was a correlation between immune cell type (or immune cell activation/inactivation) and CD200 expression in the cancer types available.
My question is can you simply do correlation analysis or something similar between the two z score columns to determine r and a p value for significance (as shown in the prism file attached) or is it not as simple as that? Or is there something else we could potentially do with the z score data?
Thank you very much in advance!