inc/qc_object.sh

phylogeny/distTree_inc/qc_object.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+THIS=`dirname $0`
+source /home/brovervv/code/cpp/bash_common.sh
+if [ $# -ne 1 ]; then
+  echo "QC an object"
+  echo "#1: file or directory with object data"
+  exit 1
+fi
+FD=$1
+
+
+exit 0

phylogeny/distTree_inc_init.sh

@@ -105,6 +105,7 @@ fi
 create_script request2dissim
 create_script request_closest
 create_script qc
+create_script qc_object
 if false; then  # deprecated
   if [ $HYBRIDNESS_MIN != 0 ]; then
   	create_script db2unhybrid

phylogeny/distTree_inc_place_seq.sh → phylogeny/distTree_inc_place.sh

@@ -5,43 +5,46 @@ if [ $# -ne 2 ]; then
   echo "Find the place of a sequence in a sequence incremental tree"
   echo "Print: <new sequence name> <tree node> <new leaf arc length> <tree node arc length>"
   echo "#1: incremental tree directory"
-  echo "#2: sequence file"
+  echo "#2: new object file or directory"
   exit 1
 fi
 INC=$1  
 QUERY=$2
 
 
-NAME=`basename $QUERY`
-ID=`head -1 $QUERY | sed 's/^>//1' | cut -f 1 -d ' '`
-if [ "$NAME" != "$ID" ]; then
-  error "File name '$NAME' must be the same as FASTA header identifier '$ID'"
-fi
+$INC/qc_object.sh $QUERY
 
 
 TMP=`mktemp`  
 #echo $TMP  
+#set -x
 
+NAME=`basename $QUERY`
 
-$THIS/../genetics/dna_closest.sh $QUERY $INC/seq.fa | grep -vw "$NAME" | sed 's/$/ '$NAME'/1' > $TMP.request
+section "Finding closest objects ..."
+$INC/request_closest.sh $NAME `dirname $QUERY` | grep -vw "$NAME" | sed 's/$/ '$NAME'/1' > $TMP.request
 
+section "Fitting to the tree ..."
 cp /dev/null $TMP.dissim
 echo "FAIL" > $TMP.leaf
 VARIANCE=`cat $INC/variance`
 while [ -s $TMP.request ]; do
-  $INC/request2dissim.sh $TMP.request $QUERY $TMP.dissim-add $TMP.log &> /dev/null
+  printf "."
+  $INC/request2dissim.sh $TMP.request $QUERY $TMP.dissim-add $TMP.log  &> /dev/null
   rm $TMP.request
   set +o errexit
   grep -vwi "nan" $TMP.dissim-add > $TMP.dissim-add1
   set -o errexit
   if [ ! -s $TMP.dissim-add1 ]; then
     wc -l $TMP.dissim-add
-    echo -e "${YELLOW}Incomparable sequences${NOCOLOR}"
+    echo -e "${YELLOW}Incomparable objects${NOCOLOR}"
     break
   fi
   cat $TMP.dissim-add1 >> $TMP.dissim
-  $THIS/../phylogeny/distTree_new  $INC/tree.released  -variance $VARIANCE  -name $NAME  -dissim $TMP.dissim  -request $TMP.request  -leaf $TMP.leaf
+  $THIS/../phylogeny/distTree_new  $INC/tree  -variance $VARIANCE  -name $NAME  -dissim $TMP.dissim  -request $TMP.request  -leaf $TMP.leaf
 done
+echo ""
+echo ""
 
 cat $TMP.leaf | cut -f 1,2,3,4
 

version.inc

@@ -1,6 +1,7 @@
-#define VERSION "1.4.2"
+#define VERSION "1.5.1"
 
 
+// 1.5.1  01/31/2021  inc/request_closest.sh has 2 parameters; inc/qc_object.sh
 // 1.4.2  01/28/2021  distTree_inc_new_log.sh
 // 1.4.1  01/24/2021  inc/{large,request_closest_sql} /dev/null or absent
 // 1.3.2  01/10/2021  GeneMark2CDS prints incomplete and ambiguous proteins as well