I want to use data from TCGA available on cbioportal to perform some statistic analysis (e.g: correlation between mRNA of a certain gene and survival data, correlation between the mRNA levels of two genes, etc). Now, in that database only Z scores are shown, both for mRNA levels and for protein levels (when available). I wanted to do a correlation test between the protein Z score of a certain gene and the mRNA Z score of a possible target of the first gene.
I wanted to check whether it's statistically correct to use the Z scores to perform this analysis, a priori it seemed quite logical to me but I wanted to make sure since I'm not really an expert in statistics.
Thanks!
Emilia Baffo
Its perfectly fine to use it. Another thing, just go a little further in the correlation tab of cbioportal, you will find a correlation table which is whole trasncriptome correlations of your queried gene. just download the table and you are all done. Now just check any gene in excel from that table. Hope it helps.
Z score is a way to standardise the data around the mean values. According to me it is fine to use them for correlation.
Yes. Find the product of the z-scores by multiplying each of the pairs of z-scores (zxzy). Then sum the products (S zxzy). Finally, divide the sum of the products by the number of scores (n) to find the correlation coefficient, r. Recall that z scores have a mean of zero. You are an expert!
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools