#!/bin/bash

# get oldest folder
dir=/home/cmccabe/Desktop/index
{
  read -r -d $'\t' time && read -r -d '' filename
} < <(find "$dir" -maxdepth 1 -mindepth 1 -printf '%T+\t%P\0' | sort -z )
printf "The oldest folder is $filename and was created on $time, analysis was performed using v1.3 of the medex pipeline by $USER at $(date "+%D %r")\n" >> /home/cmccabe/Desktop/index/log

# rename bam
cd /home/cmccabe/Desktop/index/$filename
   rename 's/^([^_]+_[^_]+)_.+$/$1.bam/' *.bam
   
# rename vcf files
cd /home/cmccabe/Desktop/index/$filename
   rename 's/^([^_]+_[^_]+)_.+$/$1.vcf/' *.vcf
   
# rename .bam.bai files
cd /home/cmccabe/Desktop/index/$filename
   rename 's/^([^_]+_[^_]+)_.+$/$1.bam.bai/' *.bam.bai

# add identifier to bam and vcf
BaseDir=/home/cmccabe/Desktop/index  # search dir
TranslationFile=/home/cmccabe/s5_files/identifier/input #input
SubDir=${1:-$filename} # specific subdir

cd "$BaseDir/$SubDir" # look in this folder
printf '%s\n' *.bam *.vcf *.bam.bai | awk '
FNR == NR  {  # process all rows and columns
    if(NF == 2) # 2 columns in input
        old[$2] = $1  # old identifier
    next  # next line
}
{    prefix = substr($0, 1, length($0) - 4)
    if(prefix in old)
        printf("mv \"%s\" \"%s%s\"\n", $0, old[prefix])
    else    printf("# No translation found for \"%s\"\n", $0) #not found
}' "$TranslationFile" - # update from

IonXpress_001 MEC2
IonXpress_002 MEC3
IonXpress_003 MEV48
R_2016_10_21_09_52_37_user_S5-00580-10-Medexome

IonXpress_007 MEV21
IonXpress_008 MEV22
IonXpress_009 MEV23
R_2016_09_21_14_01_15_user_S5-00580-9-Medexome

# No translation found for "IonXpress_007.bam"
# No translation found for "IonXpress_007.bam"
# No translation found for "IonXpress_007.bam"
# No translation found for "IonXpress_008.vcf"
# No translation found for "IonXpress_008.vcf"
# No translation found for "IonXpress_008.vcf"
# No translation found for "IonXpress_009.bam.bai"
# No translation found for "IonXpress_009.bam.bai"
# No translation found for "IonXpress_009.bam.bai"

IonXpress_007.bam  >>> MEV21.bam   ---- since the IonXpress_007 in the .bam located in the subdir matches $1 of input that .bam file is updated with $2 of input
IonXpress_007.vcf >>> MEV21.vcf   ---- since the IonXpress_007 in the .vcf  located in the subdir matches $1 of input  that .vcf file is updated with $2 of  input
IonXpress_007.bam.bai >>> MEV21.bam.bai  ---- since the IonXpress_007 in the .bam.bai  located in the subdir matches $1 of input  that .bam.bai file is updated with $2 of  input
IonXpress_008.bam  >>> MEV22.bam   ---- since the IonXpress_008 in the .bam  located in the subdir matches $1 of input  that .bam file is updated with $2 of  input
IonXpress_008.vcf >>> MEV22.vcf  ---- since the IonXpress_008 in the .vcf  located in the subdir matches $1 of input  that .vcf file is updated with $2 of  input
IonXpress_008.bam.bai >>> MEV22.bam.bai   ---- since the IonXpress_008 in the .bam.bai  located in the subdir matches $1 of input  that .bam.bai file is updated with $2 of  input
IonXpress_009.bam  >>> MEV23.bam   ---  since the IonXpress_009 in the .bam  located in the subdir matches $1 of input  that .bam file is updated with $2 of  input
IonXpress_009.vcf >>> MEV23.vcf   ---  since the IonXpress_009 in the .vcf  located in the subdir matches $1 of input  that .vcf file is updated with $2 of  input
IonXpress_009.bam.bai >>> MEV23.bam.bai  ---  since the IonXpress_009 in the .bam.bai  located in the subdir matches $1 of input  that .bam.bai file is updated with $2 of  input