write.plink? about bgdata HOT 4 CLOSED

Hi Clemens,

we think that this task is better left to PLINK. For subsetting, take a look at --keep or --remove for samples and --extract or --exclude for variants. For merging, there is --merge-list.

Cheers,
Alex

from bgdata.

Alright - thanks for the fast reply!

I would like to go for a pure R workflow one day without any system dependencies, but in that case good old system() will have to do the job a little longer.

Best,
Clemens

from bgdata.

You might be interested in the write_bed function of the BGLR package then. Your matrix needs to fit into memory and it only writes a .bed file, i.e., you need to generate the .bim and .fam files manually to get a full PLINK fileset.
The snp_writeBed function of the bigsnpr package looks also promising, but I haven't tried it yet.

Cheers,
Alex

from bgdata.

Ah - thanks for these hints. To my understanding both these packages rely on loading the data fully into memory and I wonder if this can be avoided for merging and subsetting. Your approach looks very promising in that regard.

Independent of this specific application bigsnpr is pretty interesting. Thanks for making me aware!

from bgdata.