Pre-training data for 1000 Genomes Project

[cbirchsy]

Hi,

I was wondering if you could share the pre-training data for the 1000G model and/or share more details on the preprocessing steps to generate the FASTA sequences?

Many thanks,
Callum

[JavierMenRev]

Hi @cbirchsy --

Sorry for the late reply. We used bcftools consensus function to convert the VCF genotypes into FASTA sequences. Note that for each individual we extracted two sequences. That is, after selecting a given individual and a 6kb region along their genome, we used the -I -H 1pIu and -I -H 2pIu flags when calling the function.

Let me know if this helps.