search

Joker-Jerome / UTMOST

UTMOST (unified test for molecular signatures) is a method for cross-tissue gene expression imputation for transcriptome-wide association analyses.
32 stars 16 forks source link

How to interpritate Output? #11

Open Ludmila6114 opened 3 years ago

Ludmila6114 commented 3 years ago

Well, i use my GWAS sumstat and have only tissues related to brain in the first output. (in "for" loop i have all tissues, but not all in output) Does it mean that this disease contribute only in brain tissues? In the second test, i don't understand how to use test_score ? In my dataframe i've chosen only genes with p values less than 0.05/17290. Can i stop here saying that this genes are significant ?

Thanks a lot! Young bioinf-lady

Joker-Jerome commented 3 years ago

Did you use all the tissues or just brain tissues? The test score is actually coming from the GBJ test and the way you chose significant genes should be correct.

Ludmila6114 commented 3 years ago

i use all tissues, but as the result of the first test i got only subset of tissues. I hope it is because i got only 'damaged' tissues. yes? Thank you!

Joker-Jerome commented 3 years ago

What do you mean by "damaged" tissues? I think you might need to recalculate the joint covariance by those brain tissues.

Ludmila6114 commented 3 years ago

@Joker-Jerome Thank you, everything is clear now, i got only subset of tissues because of memory problem!

The last think i wanted to ask is about "test score" as output of joint test. What is biological sense of this value? Does it represent smth in a context of gene expression level?

If yes, is it possible to say if the expression of significant genes increased or decreased and how much?

beibeixiaomu commented 2 years ago

@Joker-Jerome Thank you very much for the powerful approach! I have the same question with Ludmila6114. What is biological sense of “test score ” ? Does it mean the gene expression level increased in cases when test score >0, decreasd in cases when test score<0? Looking forward to your apply.