fixing ATCG filtering

DOH-JDJ0303 · Apr 27, 2024 · 2d4f9ec · 2d4f9ec
1 parent d6067d9
commit 2d4f9ec
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/bin/input-qc.sh b/bin/input-qc.sh
@@ -23,7 +23,7 @@ cat ${fasta} | sed 's/>.*$/@&@/g' | tr -d '\n' | tr '@' '\n' | grep -v '>' | tai
 cat seqs | sort | uniq > f1
 
 #---- FILTER 2: REMOVE SEQUENCES WITH AMBIGUOUS BASES ----#
-cat f1 | grep -vE 'R|Y|M|K|S|W|H|B|V|D|N' > f2
+cat f1 | grep -E '^[ATCG]+$' > f2
 if [ ! -s f2 ]
 then
     echo "Error: All sequences had ambiguous bases!" && exit 1

diff --git a/modules/local/bind-clusters.nf b/modules/local/bind-clusters.nf
@@ -20,7 +20,7 @@ process BIND_CLUSTERS {
     if [ -f "${prefix}-looseends.csv " ]
     then
         max_n=\$(cat *main.csv | cut -f 4 -d ',' | sort -rn)
-        mv ${prefix}-looseends.csv tmp && cat tmp | grep -v "\$header" | tr ',' '\t' | awk -v max=\${max_n} -v OFS=',' '{print $1,$2,$3,$4+max}' > ${prefix}-looseends.csv
+        mv ${prefix}-looseends.csv tmp && cat tmp | grep -v "\$header" | tr ',' '\t' | awk -v max=\${max_n} -v OFS=',' '{print \$1,\$2,\$3,\$4+max}' > ${prefix}-looseends.csv
         rm tmp
     fi
     echo \$header > ${prefix}-clusters.csv