Allow other ancestries than EUR and trans-ancestries

[sinarueeger]

Background

The file PRS_data.txt needed as input to the wdl pipeline contains examples with GWAS summary statistics of European and trans-ancestry origin.

This information is passed on to PRScs, which pulls out the corresponding samples from 1KG.

In the future, we may use summary statistics coming from other ancestries, such as East Asia or Africa, or a mix thereof.

Todo

Implement:

1. 1KG ancestries additionally to EUR: SAS, EAS, AFR, AMR.
2. Combinations of 1KG ancestries (see comment below) additionally to trans-ancestry: e.g. EUR-EAS.

Note

I assume that trans-ancestry just takes all of 1KG. But I am not sure what the best way is to implement a mix of 1KG ancestries. Two problems:

• There is a variance-bias trade-off: more samples = smaller variance, but higher bias. Something to consider.
• Studies containing European and East Asian samples might not share 50%-50% of samples. Hence EUR-EAS mix should take this into account (but again, variance-bias trade-off).

[piotor87]

This is a similar issue than the build one. ATM the ancestry field is ignored altogether and the European reference panel is automatically included. What i need to do is:

• include the ancestry field in the sumstats preprocessing
• add a Map file that localizes the ancestry data based on the ancestry string