grid_min --> pairs2dissim.grid, object2closest.grid

ncbi · Mar 31, 2021 · 401787c · 401787c
1 parent 74ee304
commit 401787c
Show file tree

Hide file tree

Showing 9 changed files with 32 additions and 24 deletions.
diff --git a/phylogeny/distTree_inc_add.sh b/phylogeny/distTree_inc_add.sh
@@ -15,7 +15,8 @@ PLACEMENT=$3
 
 
 # Cf. distTree_inc_new.sh
-GRID_MIN=`cat $INC/grid_min`
+GRID_MIN=`cat $INC/pairs2dissim.grid`
+SEARCH_GRID_MIN=`cat $INC/object2closest.grid`  
 QC=""  # -qc  
 VARIANCE=`cat $INC/variance`
 
@@ -64,15 +65,14 @@ if [ $N -gt 0 ]; then
   rm -rf $INC/log/
   mkdir $INC/log
   $THIS/../trav  $INC/search "touch $INC/log/%f" 
-  SEARCH_GRID_MIN=$(( $GRID_MIN / 100 ))  # PAR
   GRID=0
   if [ $N -lt $SEARCH_GRID_MIN ]; then
     $THIS/../trav  -step 1  $INC/search "$THIS/distTree_inc_search_init.sh $INC %f" 
   else
     GRID=1
     $THIS/../grid_wait.sh 1
     UL1=""
-    if [ -e $INC/request_closest_sql ]; then
+    if [ -e $INC/object2closest.sql ]; then
       UL1=",ul1=30"
     fi
     $THIS/../trav  -step 1  $INC/search "$QSUB_5$UL1  -N j%n  %Q$THIS/distTree_inc_search_init.sh $INC %f%Q > /dev/null" 

diff --git a/phylogeny/distTree_inc_complete.sh b/phylogeny/distTree_inc_complete.sh
@@ -12,17 +12,18 @@ INC=$1
 OBJS=$2
 
 
+#set -x
+
+
 #if false; then  
 section "QC $INC/"
 if [ -s $INC/tree ]; then
-  echo "$INC/tree must be empty"
-  exit 1
+  error "$INC/tree must be empty"
 fi
 
 N=`distTree_inc_new_list.sh $INC | wc -l`
 if [ $N -gt 0 ]; then
-  echo "$INC/new/ must be empty"
-  exit 1
+  error "$INC/new/ must be empty"
 fi
 
 sort -c $OBJS

diff --git a/phylogeny/distTree_inc_init.sh b/phylogeny/distTree_inc_init.sh
@@ -4,14 +4,14 @@ source $THIS/../bash_common.sh
 if [ $# -ne 13 ]; then
   echo "Initialize an incremental distance tree directory"
   echo "#1: output directory"
-  echo "#2: grid_min (> 0): min. number of dissimilarity requests to use GRID"
+  echo "#2: min. number of dissimilarity requests to use GRID (> 0)"
   echo "#3: variance: lin|linExp|..."
   echo "#4: hybridness_min (> 1); 0 - no hybrids"
   echo "#5: dissim_boundary (> 0 or NAN)"
   echo "#6: genogroup_barrier (> 0 or NAN)"
   echo "#7: complete path to a 'phenotype' directory or ''"
   echo "#8: large (0/1): 1 - files in #1/new/ and #7/ are grouped into subdirectories named file2hash(<file name>)"
-  echo "#9: request_closest_sql (0/1): 1 - request_closest.sh queries an SQL database and the number of concurrent connections is restricted to 30"
+  echo "#9: object2closest.sql (0/1): 1 - object2closest.sh queries an SQL database and the number of concurrent connections is restricted to 30"
   echo "#10: SQL server name or ''"
   echo "#11: database name on the SQL server or ''"
   echo "#12: complete path to a directory for bulk insert into the database or ''"
@@ -71,7 +71,8 @@ echo "1" > $INC/version
 
 mkdir $INC/hist
 
-echo $GRID_MIN > $INC/grid_min
+echo $GRID_MIN > $INC/object2closest.grid
+echo $GRID_MIN > $INC/pairs2dissim.grid
 touch $INC/runlog
 
 echo "$VARIANCE"          > $INC/variance
@@ -86,7 +87,7 @@ if [ $LARGE == 1 ]; then
 fi
 
 if [ $REQUEST_CLOSEST_SQL == 1 ]; then
-  touch $INC/request_closest_sql
+  touch $INC/object2closest.sql
 fi
 
 
@@ -102,8 +103,8 @@ create_script objects_in_tree
 if [ $SERVER ]; then
   create_script outlier2db
 fi
-create_script request2dissim
-create_script request_closest
+create_script pairs2dissim
+create_script object2closest
 create_script qc
 create_script qc_object
 if false; then  # deprecated

diff --git a/phylogeny/distTree_inc_new.sh b/phylogeny/distTree_inc_new.sh
@@ -18,7 +18,9 @@ $THIS/../check_tmp.sh
 SEED=1  # >= 1
 
 
-GRID_MIN=`cat $INC/grid_min`
+GRID_MIN=`cat $INC/pairs2dissim.grid`
+SEARCH_GRID_MIN=`cat $INC/object2closest.grid`  
+  # was: $GRID_MIN / 100
 QC=""  # -qc  
 RATE=0.015   # PAR
 VARIANCE=`cat $INC/variance`
@@ -87,15 +89,15 @@ if [ $N -gt 0 ]; then
   rm -rf $INC/log/
   mkdir $INC/log
   $THIS/../trav  $INC/search "touch $INC/log/%f" 
-  SEARCH_GRID_MIN=$(( $GRID_MIN / 100 ))  # PAR
   GRID=0
   if [ $N -lt $SEARCH_GRID_MIN ]; then
     $THIS/../trav  -step 1  $INC/search "$THIS/distTree_inc_search_init.sh $INC %f" 
   else
     GRID=1
     $THIS/../grid_wait.sh 1
     UL1=""
-    if [ -e $INC/request_closest_sql ]; then
+    if [ -e $INC/object2closest.sql ]; then
+      # PAR
       UL1=",ul1=30"
     fi
     $THIS/../trav  -step 1  $INC/search "$QSUB_5$UL1  -N j%n  %Q$THIS/distTree_inc_search_init.sh $INC %f%Q > /dev/null" 

diff --git a/phylogeny/distTree_inc_place.sh b/phylogeny/distTree_inc_place.sh
@@ -28,15 +28,15 @@ if [ -s $TMP.grep ]; then
 fi
 
 section "Finding closest objects"
-$INC/request_closest.sh $NAME `dirname $QUERY` | grep -vw "^${NAME}$" | sed 's/$/\t'$NAME'/1' > $TMP.request
+$INC/object2closest.sh $NAME `dirname $QUERY` | grep -vw "^${NAME}$" | sed 's/$/\t'$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
   printf "."
-  $INC/request2dissim.sh $TMP.request $QUERY $TMP.dissim-add $TMP.log  &> /dev/null
+  $INC/pairs2dissim.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

diff --git a/phylogeny/distTree_inc_request.sh b/phylogeny/distTree_inc_request.sh
@@ -21,7 +21,7 @@ DIR=$INC/search/$OBJ
 LOG=$INC/log/$OBJ
 
 
-$INC/request2dissim.sh $DIR/request "" $DIR/dissim.add $LOG &> $LOG.request2dissim
+$INC/pairs2dissim.sh $DIR/request "" $DIR/dissim.add $LOG &> $LOG.pairs2dissim
 if [ ! -s $DIR/dissim.add ]; then
   echo "Empty $DIR/dissim.add" > $LOG
   exit 1

diff --git a/phylogeny/distTree_inc_request2dissim.sh b/phylogeny/distTree_inc_request2dissim.sh
@@ -15,9 +15,9 @@ OUT=$3
 
 N=`cat $REQ | wc -l`
 echo "$N $REQ"
-GRID_MIN=`cat $INC/grid_min`
+GRID_MIN=`cat $INC/pairs2dissim.grid`
 if [ $N -le $GRID_MIN ]; then 
-  $INC/request2dissim.sh $REQ "" $OUT $OUT.log > /dev/null
+  $INC/pairs2dissim.sh $REQ "" $OUT $OUT.log > /dev/null
   if [ -e $OUT.log ]; then
     if [ -s $OUT.log ]; then
       head $OUT.log
@@ -41,7 +41,7 @@ else
 		  wc -l $INC/dr.list
 		  mkdir $INC/dr.log
 		  $THIS/../grid_wait.sh 1
-  		$THIS/../trav  -step 1  $INC/dr.list "$QSUB_5 -N j%f %q$INC/request2dissim.sh $INC/dr/%f %Q%Q $INC/dr.out/%f $INC/dr.log/%f%q > /dev/null"		
+  		$THIS/../trav  -step 1  $INC/dr.list "$QSUB_5 -N j%f %q$INC/pairs2dissim.sh $INC/dr/%f %Q%Q $INC/dr.out/%f $INC/dr.log/%f%q > /dev/null"		
   		$THIS/../qstat_wait.sh 3000 1	 
 		  ls $INC/dr.log > $INC/dr.bad
 		  ls $INC/dr.out > $INC/dr.good

diff --git a/phylogeny/distTree_inc_search_init.sh b/phylogeny/distTree_inc_search_init.sh
@@ -16,7 +16,7 @@ DIR=$INC/search/$OBJ
 
 while true; do
   set +o errexit
-	$INC/request_closest.sh $OBJ "" > $DIR/request.raw
+	$INC/object2closest.sh $OBJ "" > $DIR/request.raw
   S=$?
   set -o errexit
 	if [ $S == 0 ]; then

diff --git a/version.inc b/version.inc
@@ -1,6 +1,10 @@
-#define VERSION "1.5.10"
+#define VERSION "1.6.1"
 
 
+// 1.6.1  03/31/2021  inc/: request2dissim.sh -> pairs2dissim.sh
+//                          request_closest.sh -> object2closest.sh
+//                          request_closest_sql -> object2closest.sql
+//                          grid_min -> pairs2dissim.grid and object2closest.grid
 // 1.5.10 03/25/2021  inc/request_closest.sh $OBJ may return $OBJ
 // 1.5.9  03/24/2021  NewLeaf::saveRequest() prints empty list of requests if location.indiscernible()
 // 1.5.8  03/22/2021  makeDistTree: quality calculation uses threads