This query is regarding imputing indels in 4x coverage WGS data using GLIMPSE1. I already imputed the SNPs in my dataset using GLIMPSE1 following the tutorial. GLIMPSE1 was used as my reference data set is smaller (661) than my target dataset (~2000) and wanted to use the joint model approach.
Imputation worked well. However, I also want to impute the indels in the dataset and while going through the Q&As, I found this approach [https://github.com/odelaneau/GLIMPSE/issues/104], where it suggests imputing indels without calling them using a variant caller. So, my approach is to:
create the reference panel vcf.gz and tsv.gz with both SNPs and indels (I'm using the GRCh37 1000Genome African samples =661)
compute GLs only for SNP positions using BCFtools
imputation with GLIMPSE_phase using --impute-reference-only-variants
run GLIMPSe_ligate and GLIMPSE_sample
Or should I just create the reference with only indels and directly perform GLIMPSE_phase using --impute-reference-only-variants followed by GLIMPSE_ligate? Or would you have better suggestion to include indels since I already have imputed SNPs? Greatly appreciate your thoughts on this.
Hi Simone,
This query is regarding imputing indels in 4x coverage WGS data using GLIMPSE1. I already imputed the SNPs in my dataset using GLIMPSE1 following the tutorial. GLIMPSE1 was used as my reference data set is smaller (661) than my target dataset (~2000) and wanted to use the joint model approach.
Imputation worked well. However, I also want to impute the indels in the dataset and while going through the Q&As, I found this approach [https://github.com/odelaneau/GLIMPSE/issues/104], where it suggests imputing indels without calling them using a variant caller. So, my approach is to:
Or should I just create the reference with only indels and directly perform GLIMPSE_phase using --impute-reference-only-variants followed by GLIMPSE_ligate? Or would you have better suggestion to include indels since I already have imputed SNPs? Greatly appreciate your thoughts on this.
Thanks a lot! Best Rangi