format.subs

#!/bin/bash

export SomaLogic=/scratch/jhz22/SomaLogic
export bs=185

function FHS()
{
awk 'NR==(b-1)*bs+1, NR==b*bs {
   cmd=sprintf("gunzip -c %s/FHS/X_%s.txt.gz | \
       awk -vFS=\",\" -vOFS=\"\\t\" -f doc/FHS.awk | \
       sort -k2,2n -k3,3n | \
       awk -f doc/order.awk | \
       gzip -f > %s/FHS/FHS.%s.txt.gz",box,$1,sumstats,$2)
   system(cmd)
}' b=$SLURM_ARRAY_TASK_ID bs=$bs box=$SomaLogic/box sumstats=$SomaLogic/sumstats $SomaLogic/sumstats/FHS.list
}

function KORA()
{
awk 'NR==(b-1)*bs+1, NR==b*bs {
    cmd=sprintf("gunzip -c %s/KORA/KORA_pGWAS.%s.assoc.linear.gz | \
        sort -k2,2 | \
        join -j2 - %s/KORA.bim | \
        awk -vOFS=\"\\t\" -f doc/KORA.awk | \
        sort -k2,2n -k3,3n | \
        awk -f doc/order.awk | \
        gzip -f > %s/KORA/KORA.%s.txt.gz",box,$1,sumstats,sumstats,$1)
    system(cmd)
}' b=$SLURM_ARRAY_TASK_ID bs=$bs box=$SomaLogic/box sumstats=$SomaLogic/sumstats $SomaLogic/sumstats/KORA.list
}

function Malmo()
{
awk 'NR==(b-1)*bs+1, NR==b*bs {
    cmd=sprintf("gunzip -c %s/Malmo/zln%s_summary.csv.gz | \
        awk -vFS=\",\" -vOFS=\"\\t\" -f doc/Malmo.awk | \
        sort -k2,2n -k3,3n | \
        awk -f doc/order.awk | \
        gzip -f > %s/Malmo/Malmo.%s.txt.gz",box,$1,sumstats,$2)
    system(cmd)
}' b=$SLURM_ARRAY_TASK_ID bs=$bs box=$SomaLogic/box sumstats=$SomaLogic/sumstats $SomaLogic/sumstats/Malmo.list
}

function QMDiab()
{
awk 'NR==(b-1)*bs+1, NR==b*bs {
    cmd=sprintf("gunzip -c %s/QMDiab/PGWAS_Results/QMDiab_pGWAS.%s.assoc.linear.gz | \
        sort -k2,2 | \
        join -j2 - %s/QMDiab.bim | \
        awk -vOFS=\"\\t\" -f doc/QMDiab.awk | \
        sort -k2,2n -k3,3n | \
        awk -f doc/order.awk | \
        gzip -f > %s/QMDiab/QMDiab.%s.txt.gz",box,$1,sumstats,sumstats,$1)
    system(cmd)
}' b=$SLURM_ARRAY_TASK_ID bs=$bs box=$SomaLogic/box sumstats=$SomaLogic/sumstats $SomaLogic/sumstats/QMDiab.list
}

export TMPDIR=/scratch/jhz22/SomaLogic/sumstats/$1
$1