How to convert the result predicted by PrediXcan to the real TPM value

[xinyu liang]

Hello,

I'm using PrediXcan to predict gene expression value from 1000 genome dataset, and the result will be used for other analysis which needs TPM value. 

When I make correlation between the predictive results and GEUVADIS expression data, the Pearson's correlation is near to zero. I want to know either it is caused by somthing wrong in my analysis or I need to do some conversion. Is there any method to achieve this conversion?

Thanks, I really need some help.

[Hae Kyung Im]

You should be getting a range of correlations (squared shown in the figure), so there is something wrong with they way you are computing the prediction.

https://pubmed.ncbi.nlm.nih.gov/26258848/#&gid=article-figures&pid=figure-3-uid-2

To help you, we would need a more detailed description on how you are computing the correlations. Exact command line options and heads of input files, etc.

Haky

--
You received this message because you are subscribed to the Google Groups "PrediXcan/MetaXcan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to predixcanmetax...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/predixcanmetaxcan/8912e93b-6f17-4c88-b11b-6afa8bdb784bn%40googlegroups.com.