search

hakyim / TO-DELETE-PrediXcan

Code for the in-dev PrediXcan Project
MIT License
28 stars 83 forks source link

Population difference #33

Open HongyanR opened 5 years ago

HongyanR commented 5 years ago

Hi, a quick question: Can I use the trained .db files to predict expression level in the Chinese population? I noticed all your files are named "...imputed_europeans".

hakyim commented 5 years ago

Dear Hongyan

as the file names suggest, the prediction models were training using predominantly European samples. We have performed some comparisons and found that the cross population is possible although we do lose some performance. This paper will give you a better sense of how much performance is lost https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1007586

At the moment, we don't have GTEx scale data in other populations so my advice would be to use this and if possible also use a separate population specific model.

I hope this helps Haky

On Mon, Mar 4, 2019 at 8:55 PM HongyanR notifications@github.com wrote:

Hi, a quick question: Can I use the trained .db files to predict expression level in the Chinese population? I noticed all your files are named "...imputed_europeans".

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/hakyim/PrediXcan/issues/33, or mute the thread https://github.com/notifications/unsubscribe-auth/AC2ouScOvbqykekDV8zSI9tO5RX_Dvblks5vTdy5gaJpZM4bdpmq .