From 370339bb119e5368fc21d9811009dfc71522d3b5 Mon Sep 17 00:00:00 2001
From: Vyacheslav Brover <vyacheslav.brover@nih.gov>
Date: Wed, 19 Feb 2025 12:50:59 -0500
Subject: [PATCH] #3: min. complexity

---
 genetics/marker2qual.sh | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/genetics/marker2qual.sh b/genetics/marker2qual.sh
index b1028f76..8a284d45 100755
--- a/genetics/marker2qual.sh
+++ b/genetics/marker2qual.sh
@@ -1,16 +1,18 @@
 #!/bin/bash --noprofile
 THIS=$( dirname $0 )
 source $THIS/../bash_common.sh
-if [ $# -ne 3 ]; then
+if [ $# -ne 4 ]; then
   echo "Print a good quality subset of eukaryotic marker proteins created by tblastn2marker_euk.sh"
   echo "#1: marker proteins (FASTA)"
   echo "#2: min. score to length ratio"
-  echo "#3: output uniKernel file | ''"
+  echo "#3: min. complexity"
+  echo "#4: output uniKernel file | ''"
   exit 1
 fi
 M=$1
 T=$2
-UNI=$3
+COMPL=$3
+UNI=$4
 
 
 TMP=$( mktemp )
@@ -18,13 +20,13 @@ TMP=$( mktemp )
 
 grep '^>' $M | cut -f 1,7 -d ' '| sed 's/^>//1' | sed 's/ score=/\t/1' > $TMP.score
 $THIS/fasta2len $M > $TMP.len
-paste $TMP.len $TMP.score | awk -F '\t' '{OFS="\t"; print $1, $4/$2};' > $TMP.stat
+paste $TMP.len $TMP.score | awk -F '\t' '{OFS="\t"; print $1, $4 / $2};' > $TMP.stat
 if [ "$UNI" ]; then
   $THIS/../dm/conversion/cols2dm.sh $TMP.stat 0 5 1 > $TMP.dm
   $THIS/../dm/uniKernel $TMP "V2" -qc > $UNI
 fi
 awk '$2 > '$T $TMP.stat | cut -f 1 > $TMP.list
-$THIS/filterFasta $M  -aa  -target $TMP.list  -len_min 20  -complexity_min 3
+$THIS/filterFasta $M  -aa  -target $TMP.list  -len_min 20  -complexity_min $COMPL
 
 
 rm $TMP*