search

gusevlab / fusion_twas

methods for functional summary-based imputation
http://gusevlab.org/projects/fusion/
GNU General Public License v3.0
75 stars 45 forks source link

Imputing Individual Sample Expression Values #9

Closed jgockley62 closed 5 years ago

jgockley62 commented 5 years ago

I was wondering what the best way to export imputed gene expression values for each genotype into a matrix to analyze. Is that something I can change FUSION.assoc_test.R to do this?

sashagusev commented 5 years ago

Can you clarify what kind of output you're looking for? Each of the weight files can be loaded in R and contains a matrix of expression weights. You can also use the make_score function (https://github.com/gusevlab/fusion_twas/blob/master/utils/make_score.R) to produce predictor scores that can be loaded with the plink --score function to do individual-level prediction/imputation.

jgockley62 commented 5 years ago

Cool I can use the make score on a gene by gene case and combine post-run. But I was also wondering don't some of the models use intercept values that don't seem to be used used by plink --score ?

sashagusev commented 5 years ago

Yes, you can make_score on all the genes, predict each into your data, and then analyse together in multi-variate model. The models always remove the intercept (i.e. center the phenotype and genotype) so that information does not get used.