Adding parser, mutation origin, muation type, and group identificatio…

…n scripts for SNV files.
asw0049 · Oct 2, 2022 · b8a5127 · b8a5127
1 parent 1e2d11e
commit b8a5127
Show file tree

Hide file tree

Showing 7 changed files with 114 additions and 0 deletions.
diff --git a/SNP_positionFreqGroup.sh b/SNP_positionFreqGroup.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+###This script reports frequenct of mutations at certain positions by group.
+##Dependencies: groupType3.awk, SNV_tstv_rosID.awk
+tissue=muscle
+
+
+cat ${tissue}MTsnps.txt | awk -F";" '{print $8}'| sed 's/_/;/g'| awk -F";" '{print ";"$1}'> ${tissue}_SampleTags.txt
+paste -d'\0' ${tissue}MTsnps.txt ${tissue}_SampleTags.txt> ${tissue}MTsnps2.txt
+cat ${tissue}MTsnps2.txt | awk -F";" -f groupType3.awk > ${tissue}MTsnps_labelled.txt
+grep -w "Middle" ${tissue}MTsnps_labelled.txt | awk -F";" '{print $2";"$3";"$7";"$11}' | sort -n| uniq -c| sort -r | awk -F" " '{print $1";"$2}'| awk -F";" -f SNV_tstv_rosID.awk >${tissue}.MTsnpsPos.middle.txt
+grep -w "Aged" ${tissue}MTsnps_labelled.txt | awk -F";" '{print $2";"$3";"$7";"$11}' | sort -n| uniq -c| sort -r | awk -F" " '{print $1";"$2}'| awk -F";" -f SNV_tstv_rosID.awk >${tissue}.MTsnpsPos.aged.txt
+grep -w "NR" ${tissue}MTsnps_labelled.txt | awk -F";" '{print $2";"$3";"$7";"$11}' | sort -n| uniq -c| sort -r | awk -F" " '{print $1";"$2}'| awk -F";" -f SNV_tstv_rosID.awk > ${tissue}.MTsnpsPos.nr.txt
+grep -w "R" ${tissue}MTsnps_labelled.txt | awk -F";" '{print $2";"$3";"$7";"$11}' | sort -n| uniq -c| sort -r | awk -F" " '{print $1";"$2}'| awk -F";" -f SNV_tstv_rosID.awk > ${tissue}.MTsnpsPos.r.txt
+
+exit
+
+
diff --git a/SNV_tstv_rosID.awk b/SNV_tstv_rosID.awk
@@ -0,0 +1,10 @@
+{
+if (($3 == "A-T") || ($3 == "A-C") || ($3 == "T-A") || ($3 == "T-G") || ($3 == "G-C") || ($3 == "C-G")) SNVType="Other Transversion";
+else if (($3 == "A-G") || ($3 == "G-A") || ($3 == "T-C") || ($3 == "C-T")) SNVType="Replication Error Transition";
+else if (($3 == "G-T") || ($3 == "C-A")) SNVType="ROS-Mediated Transversion";
+else if ($3 ~ /[[:alpha:]]-[[:alpha:]]{2}+/) SNVType="Replication Error Insertion";
+else if ($3 ~ /[[:alpha:]]{2}+-[[:alpha:]]/) SNVType="Replication Error Deletion";
+print ($0)";"SNVType
+if (($3 != "A-T")||($3 != "A-C")||($3 != "A-G")||($3 != "G-C")||($3 != "G-A")||($3 != "G-T")||($3 != "C-A")||($3 != "C-T")||($3 !="C-G")||($3 != /[[:alpha:]]-[[:alpha:]]{2}+/)||($3 != /[[:alpha:]]{2}+-[[:alpha:]]/)) SNVType="Other";
+}
+
diff --git a/geneCountParser.sh b/geneCountParser.sh
@@ -0,0 +1,37 @@
+#!/bin/bash
+##Script Variables
+input_file=muscle.sample.list.txt
+snp_file=muscleMTsnps.txt
+gene_file=muscle.gene.list.txt
+tissue=muscle
+#####Setup Environment
+#Dependency Scripts
+##groupType2.awk
+###Commands
+
+#cat $snp_file | awk -F";" '{print $8}'|sort|uniq > ${tissue}.sample.list.txt
+
+cat $snp_file | awk -F";" '{print $7}'|sort|uniq > ${tissue}.gene.list.txt
+
+if test -f "$input_file"; then
+    while read F  ; do
+        echo $F
+        declare -a sample_list=($F)
+        for sample in ${sample_list[@]}; do
+             cat $snp_file | grep ${sample}> ${sample}_${tissue}Muts.txt
+             while read A ; do
+                 echo $A
+                 declare -a gene_list=($A)
+                 for gene in ${gene_list[@]}; do
+                     cat ${sample}_${tissue}Muts.txt| grep $gene | awk -F";" -v g=$gene '{if ($7 == g ) { count += 1; }} END { print $7";"$8";" count}' > ${sample}_${gene}.mut.txt
+                 done
+             done <$gene_file
+        done
+    done <$input_file
+    cat *.mut.txt | grep -v ";;" | sed 's/_/;/g'| awk -F";" -f groupType2.awk >${tissue}GeneMutCount.txt
+    rm *.mut.txt
+else
+    echo "$input_file not found. Aborting!!!!"
+fi
+exit
+
diff --git a/groupType.awk b/groupType.awk
@@ -0,0 +1,7 @@
+{
+if (($1 >= 1) && ($1 <= 16) || ($1 >= 61) && ($1 <= 76) ||  ($1 >= 121) && ($1 <= 136)) groupType="Middle";
+if (($1 >= 17) && ($1 <= 25) || ($1 >= 77) && ($1 <= 85) ||  ($1 >= 137) && ($1 <= 145)) groupType="Aged";
+if (($1 >= 26) && ($1 <= 47) || ($1 >= 86) && ($1 <= 107) ||  ($1 >= 146) && ($1 <= 167)) groupType="NR";
+if (($1 >= 48) && ($1 <= 60) || ($1 >= 108) && ($1 <= 120) ||  ($1 >= 168) && ($1 <= 180)) groupType="R";
+print $1";"$4";"$5";"groupType;
+}
diff --git a/groupType2.awk b/groupType2.awk
@@ -0,0 +1,7 @@
+{
+if (($2 >= 1) && ($2 <= 16) || ($2 >= 61) && ($2 <= 76) ||  ($2 >= 121) && ($2 <= 136)) groupType="Middle";
+if (($2 >= 17) && ($2 <= 25) || ($2 >= 77) && ($2 <= 85) ||  ($2 >= 137) && ($2 <= 145)) groupType="Aged";
+if (($2 >= 26) && ($2 <= 47) || ($2 >= 86) && ($2 <= 107) ||  ($2 >= 146) && ($2 <= 167)) groupType="NR";
+if (($2 >= 48) && ($2 <= 60) || ($2 >= 108) && ($2 <= 120) ||  ($2 >= 168) && ($2 <= 180)) groupType="R";
+print ($0)";"groupType;
+}
diff --git a/groupType3.awk b/groupType3.awk
@@ -0,0 +1,8 @@
+{
+if (($10 >= 1) && ($10 <= 16) || ($10 >= 61) && ($10 <= 76) ||  ($10 >= 121) && ($10 <= 136)) groupType="Middle";
+if (($10 >= 17) && ($10 <= 25) || ($10 >= 77) && ($10 <= 85) ||  ($10 >= 137) && ($10 <= 145)) groupType="Aged";
+if (($10 >= 26) && ($10 <= 47) || ($10 >= 86) && ($10 <= 107) ||  ($10 >= 146) && ($10 <= 167)) groupType="NR";
+if (($10 >= 48) && ($10 <= 60) || ($10 >= 108) && ($10 <= 120) ||  ($10 >= 168) && ($10 <= 180)) groupType="R";
+print ($0)";"groupType;
+}
+
diff --git a/mutParser.sh b/mutParser.sh
@@ -0,0 +1,28 @@
+#!/bin/bash
+##Script Variables
+input_file=muscle.sample.list.txt
+snp_file=muscleMTsnps.txt
+tissue=muscle
+##Setup Environment
+
+###Commands
+
+cat $snp_file | awk -F";" '{print $8}'|sort|uniq > ${tissue}.sample.list.txt
+
+if test -f "$input_file"; then
+    while read F  ; do
+        echo $F
+        declare -a sample_list=($F)
+        for sample in ${sample_list[@]}; do
+             cat $snp_file | grep ${sample} | grep "Transition" | awk -F";" '{if ($9 == "Transition" ) { TScount += 1; }} END { print $8";"$9";" TScount}' >> ${tissue}_mut_Count_individual.txt
+             cat $snp_file | grep ${sample} | grep "Transversion" | awk -F";" '{if ($9 == "Transversion" ) { TVcount += 1; }} END { print $8";"$9";" TVcount}' >> ${tissue}_mut_Count_individual.txt
+             cat $snp_file | grep ${sample} | grep "Insertion" | awk -F";" '{if ($9 == "Insertion" ) { Incount += 1; }} END { print $8";"$9";" Incount}' >> ${tissue}_mut_Count_individual.txt
+             cat $snp_file | grep ${sample} | grep "Deletion" | awk -F";" '{if ($9 == "Deletion" ) { Delcount += 1; }} END { print $8";"$9";" Delcount}' >> ${tissue}_mut_Count_individual.txt
+       done
+    done <$input_file
+    cat ${tissue}_mut_Count_individual.txt | grep -v ';;' | sed 's/_/;/g' | awk -F";" -f groupType.awk > ${tissue}_mut_Count_individual2.txt
+else
+    echo "$input_file not found. Aborting!!!!"
+fi
+exit
+