Introduction

The SNP array was imputed using hg19 coordinates which is conflicting with the hg38 build (used for WGS). Therefore, the array data has to be converted to hg38 before re-running all summary statistics and further analyses.

Methods

1.Access taurus through WinSCP

2.Transfer files from local directory to remote directory

/data/all/martha

3.Convert bed files to ped files

plink --bfile NImerged.impute2.chrAllX.2014-Oct-02 --recode --tab --out plink_output/NImerged.impute2.chrAllX

This has converted the .bed .bim .fam files to .ped .map files.

4. Download the following files:

LiftOverPlink python scripts: https://github.com/sritchie73/liftOverPlink LiftOver binary: http://hgdownload.cse.ucsc.edu/admin/exe/linux.x86_64/liftOver Chain files: http://hgdownload.cse.ucsc.edu/goldenPath/hg19/liftOver/hg19ToHg38.over.chain.gz

5. Create a lifted map file

python liftOverPlink-master/liftOverPlink.py --bin /data/all/martha/Data/liftOver --map plink_output/Nimerged.impute.chrAllX.map --out plink_output/Nimerged.impute_lifted --chain hg19toHg38.over.chain

Note: permissions had to be modified for the binary to work

The “lifted” map file is the product of conversion between genome builds. Typically, .map files contain chr. numbers, marker ID/SNP ID, genetic distance, and position of the SNP. However, not all SNPs can be converted between builds for various reasons. These have to be removed.

6. Remove “bad” lifted SNPs from new .map file

python liftoverPlink-master/rmBadLifts.py --map plink_output/Nimerged.impute_lifted.map --out plink_output/NI.snp.good_lifted.map --log NI.snp.bad_lifted.dat

7. Create a file that contains “unlifted” and “bad” SNPs.

cut -f 2 plink_output/NI.snp.bad_lifted.dat > plink_output/NI.snps_exclude.dat
cut -f 4 plink_output/NImerged.impute_fted.bed.unlifted | sed "/^#/d" >> plink_output/NI.snps_exclude.dat

8. Generate a .ped file that excludes the “unlifted” and “bad” SNPs.

plink --file plink_output/Nimerged.impute.chrAllX --recode --out plink_output/NI_snp_hg38 --exclude plink_output/NI.snps.exclude.dat

Typically, .ped files contain family IDs, UUID, paternal ID, maternal ID, sex, affection, and genotypes.

Output: - there are 27,310,409 variants in the original .map file - after excluding unlifted and bad SNPs: 26,055,300 SNPs remain - there are 824 duplicate variant IDs in the exclude file - total genotyping rate of the new file is 98.69%

9. Combine the “good” .ped and .map files

plink --ped plink_output/NI_snp_hg38.ped --map plink_output/NI.snp.good_lifted.map --recode --out plink_output/NI.snp.hg38_final

10. Convert ped/map files to bed/bim/fam files

plink --file plink_output/NI.snp.hg38_final --make-bed --out NI.snp.hg38_final

Converting SNP array data