new

dm/nominal2uniKernel.sh

@@ -0,0 +1,50 @@
+#!/bin/bash --noprofile
+THIS=$( dirname $0 )
+source $THIS/../bash_common.sh
+if [ $# -ne 4 ]; then
+  echo "Dependence of #2 on the other nominal columns"
+  echo "#1: tsv-file for ANOVA. Column names should be unique w.r.t. space-underscore replacement"
+  echo "#2: target column name"
+  echo "#3: uniKernel window (0 - default)"
+  echo "#4: output directory"
+  exit 1
+fi
+TSV=$1  
+TARGET="$2"
+WINDOW=$3
+OUT=$4
+
+
+TMP=$( mktemp )
+#comment $TMP
+#set -x
+
+
+head -1 $TSV | sed 's/^#//1' | tr '\t' '\n' > $TMP.header
+mapfile -t H < $TMP.header
+T=$( grep -n -x "$TARGET" $TMP.header | sed 's/:.*$//1' )
+
+I=0
+while [ $I -lt ${#H[@]} ]; do
+  if [ "${H[$I]}" != "$TARGET" ]; then
+    ATTR="${H[$I]}"
+    I1=$(( I + 1 ))
+    tail -n +2 $TSV | cut -f $I1 | sort -u > $TMP.list
+    M=$( < $TMP.list wc -l )
+    echo "$ATTR: $M"
+    mapfile -t L < $TMP.list
+    J=0
+    while [ $J -lt ${#L[@]} ]; do
+      V="${L[$J]}"
+      echo "  $V"
+      $THIS/../tsv/tsv_awk.sh $TSV '$'$I1' == "'"$V"'"' | tail -n +2 | cut -f $T > $TMP.data
+      $THIS/conversion/cols2dm.sh $TMP.data 0 6 0 > $TMP.dm
+      $THIS/uniKernel $TMP "V1" -noprogress  -window $WINDOW  > "$OUT/$ATTR-$V.uniKernel"
+      J=$(( J + 1 ))
+    done
+  fi
+  I=$(( I + 1 ))
+done
+
+
+rm $TMP*