######################################
##### UKBB GWAS QC PLINK2 script #####
######################################

gendir=mydir

# iterated over each chromosome # 

i=$1
echo "chrom-"$i

bgeni=ukb_imp_chr${i}_v3
samplei=ukb#####_imp_chr${i}_v3_s###### # sample ID

bgenf=${gendir}/${bgeni}.bgen
samplef=${gendir}/${samplei}.sample

# Participant QC
plink2 --bgen $bgenf --sample ${samplef} --keep qc_all_subjs.fam --make-bed --out chr${i}/sample_filter_${bgeni}

# HWE
plink2 --bfile chr${i}/sample_filter_${bgeni} --hwe 1e-10 --make-bed --out chr${i}/hwe_filter_${bgeni}

# MAF
plink2 --bfile chr${i}/hwe_filter_${bgeni} --maf 0.01 --make-bed --out chr${i}/maf_filter_${bgeni}

#wget  -nd  biobank.ctsu.ox.ac.uk/crystal/crystal/auxdata/ukb_imp_mfi.tgz
#tar zxvf ukb_imp_mfi.tgz

# Create InfoMaf_filter.txt files
infoi=ukb_mfi_chr${i}_v3
awk -F" " -v info=0.8 '$8 > info { print }' ${infoi}.txt > tmp0${i}
cut -f2 tmp0${i} > ${infoi}_filtered.txt
rm tmp0${i}

# INFO
plink2 --bfile chr${i}/maf_filter_${bgeni} --extract ${infoi}_filtered.txt --make-bed --out chr${i}/info_filter_${bgeni}

# CALL RATE
plink2 --bfile chr${i}/info_filter_${bgeni} --geno 0.05 --make-bed --out chr${i}/anal_${bgeni}

# clean
rm chr${i}/sample_filter_${bgeni}.*
rm chr${i}/hwe_filter_${bgeni}.*
rm chr${i}/maf_filter_${bgeni}.*
rm chr${i}/info_filter_${bgeni}.*


##########################################
##### GWAS association PLINK2 script #####
##########################################

# iterated over each chromosome # 

# specify phenotype name
s_pheno=phenotype_name

# phenotype file 
pfile=pheno_${s_pheno}

# specify output directory 
outdir=specify_outdir

i=$1
echo "chrom-"$i
bfilei=QC_gwas/anal_ukb_imp_chr${i}_v3

# plink2 association commands 

plink2 --bfile $bfilei \
       --glm omit-ref hide-covar \
       --pheno $pfile \
       --variance-standardize \
       --covar covars_${s_pheno} \
       --out ${outdir}/${i}_assoc_out_${s_pheno}