loco for chrX

[joshchiou]

Is it possible to also output loco for non-autosomal chromosomes? This would make it possible to run step 2 on chrX.
This closed issue from regenie references how they calculate the loco values for non-autosomes.

rgcgithub/regenie#353

[hrushikeshloya]

Hi, if you run step 1 with --predBetasFlag argument you'll get an additional line in your loco files (as Regenie). You could then rename your chromosome X as last autosomal chromosome number + 1 (23 in case you are using 22 chromosomes for model fitting) i.e. change the chromosome name in bim/bgen file. Running Quickdraws step 2 on the renamed file would use LOCO predictions from all chromosomes considered in step 1 (same approach as Regenie) and output summary statistics for chr X.

Let me know if this doesn't work or if this doesn't answer your query.

[joshchiou]

Thank you! I can confirm adding --predBetasFlag works exactly as you described.

One thing to note, and I'm not sure if this is due to recent changes in 0.1.5 or because of adding this new argument, but the indices for the LOCO predictions are now floats (i.e. 1.0 instead of 1).

[hrushikeshloya]

Thanks for letting me know, does this affect any analysis downstream?