lifebit-ai / gwas

GWAS pipeline using SAIGE
5 stars 2 forks source link

Allow continuous variable as PHE #39

Closed cgpu closed 4 years ago

cgpu commented 4 years ago

https://github.com/lifebit-ai/gel-gwas/blob/a13eb4eb85930d9ecd6db42053cc4763fe54e792/main.nf#L255

# quantitative
--traitType=quantitative
--invNormalize=TRUE 
# binary
--traitType=binary
mcamarad commented 4 years ago

@cgpu When you test for missingness in a GWAS is it possible to do it when you have a quantitative trait? I was trying the pipeline with qt traits and couldn't make it work because the option seems to be only for case/control ones. Found this on the ref: https://www.cog-genomics.org/plink/1.9/assoc#test_missing

This issue affects this part of the code: https://github.com/lifebit-ai/gel-gwas/blob/a84f43f084be986c9b3d7b0b1587902bb69b7f38/main.nf#L198-L208

So far, I've decided to create a switch and remove the missingness part from the qt processes which seems to work fine with qt now.

cgpu commented 4 years ago

So far, I've decided to create a switch and remove the missingness part from the qt processes which seems to work fine with qt now.

For 1st iter should be good, as discussed earlier having the test with a file that has NAs in the col that is the PHE col, and also is quantitative cols {pheno_col: {isQuantitative, hasNA}} should give us more info on expected behaviour and the possibility to catch errors.

kousathanas commented 4 years ago

for quantitative traits there is no need to test for missingness in the QC step.

cgpu commented 4 years ago

for quantitative traits there is no need to test for missingness in the QC step.

Thanks @kousathanas for verification, appreciate it!