From 2d404b5b246c8b8693d2b7e633c3efbd2bdf3d00 Mon Sep 17 00:00:00 2001
From: alain viari <alain.viari@inria.fr>
Date: Sat, 14 Nov 2015 22:13:55 +0100
Subject: [PATCH] removed need of R igraph from chlorodb/subdb

Former-commit-id: 574aace9be5804d728a877110f5f475d61644f75
Former-commit-id: 2e7ea63447643830a62f18a364327d7b396ec140
---
 TODO                                          |  5 +++
 detectors/cds/tools/chlorodb/README.txt       | 20 +++++------
 .../cds/tools/chlorodb/subdb/go_subdb.sh      | 10 ++----
 .../cds/tools/chlorodb/subdb/lib/db.cc.awk    | 31 +++++++++++++++++
 .../cds/tools/chlorodb/subdb/lib/db.cc.r      | 18 ----------
 .../chlorodb/subdb/lib/db.filter.len.awk      | 33 +++++++++++++++++++
 .../tools/chlorodb/subdb/lib/db.filter.len.r  | 19 -----------
 .../tools/chlorodb/subdb/lib/db.getlen.awk    | 10 ------
 detectors/cds/tools/lib/install.rpackages.r   |  1 -
 9 files changed, 81 insertions(+), 66 deletions(-)
 create mode 100644 detectors/cds/tools/chlorodb/subdb/lib/db.cc.awk
 delete mode 100755 detectors/cds/tools/chlorodb/subdb/lib/db.cc.r
 create mode 100644 detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.awk
 delete mode 100755 detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.r
 delete mode 100644 detectors/cds/tools/chlorodb/subdb/lib/db.getlen.awk

diff --git a/TODO b/TODO
index 66e715a..af736c1 100644
--- a/TODO
+++ b/TODO
@@ -18,3 +18,8 @@ o parameters in go_subdb.sh
 
 o models in DB_DIR
 
+o check R package igraph installation
+
+o replace R igraph for cc
+
+
diff --git a/detectors/cds/tools/chlorodb/README.txt b/detectors/cds/tools/chlorodb/README.txt
index c58508c..9adf986 100644
--- a/detectors/cds/tools/chlorodb/README.txt
+++ b/detectors/cds/tools/chlorodb/README.txt
@@ -23,17 +23,15 @@ then (after checking)
 # notes
 #
 
-this requires an installed R > 3.0.1
+calculation of models currently requires R
+(without any specific package) this will be
+replaced in the future...
 
-with the following packages
+in addition, optional graphics output (plot.models.r) 
+requires the following graphic packages :
 
-    igraph          # <- mandatory
+    grid           
+    gridExtra      
+    vcd            
+    plotrix        
 
-    grid            # <- the following are not needed
-    gridExtra       #    by scripts, but just to
-    vcd             #    produce graphics for models
-    plotrix         #    by lib/plot.models.r
-
-you can check and install them by running :
-
- lib/install.rpackages.r 
diff --git a/detectors/cds/tools/chlorodb/subdb/go_subdb.sh b/detectors/cds/tools/chlorodb/subdb/go_subdb.sh
index 0ea30fb..0d7a5b2 100755
--- a/detectors/cds/tools/chlorodb/subdb/go_subdb.sh
+++ b/detectors/cds/tools/chlorodb/subdb/go_subdb.sh
@@ -111,9 +111,7 @@ Notify "select by length"
 
 foreach f (D_$$/*.fst) 
   set nom = `basename $f:r`
-  $AwkCmd -f $LIB_DIR/db.getlen.awk $f > L_$$
-  $LIB_DIR/db.filter.len.r L_$$ $Delta |\
-    $AwkCmd '($NF == "TRUE") {print $2}' > M_$$
+  $AwkCmd -v DELTA=$Delta -f $LIB_DIR/db.filter.len.awk $f > M_$$
   $AwkCmd -v FILE=M_$$ -f $LIB_DIR/db.subdb.awk $f > E_$$/$nom.fst
   Report E_$$/$nom.fst "length_filter"
   set n = `egrep '^>' E_$$/$nom.fst | wc -l`
@@ -144,10 +142,8 @@ foreach f (E_$$/*.fst)
   
   $AwkCmd -v COVMIN=$Covmin -v PMAX=$Pmax -v IDMIN=$Idmin \
       -f $LIB_DIR/db.blastlink.awk $f.blast.out |\
-      $AwkCmd -f $LIB_DIR/db.todl.awk > G_$$
-      
-  ($LIB_DIR/db.cc.r G_$$ > $f.cc.txt) >>& db.log
-  
+  $AwkCmd -f $LIB_DIR/db.cc.awk > $f.cc.txt
+
   awk -v NAME=$nom -f $LIB_DIR/db.reportcc.awk $f.cc.txt >> $OutLog 
   
   $AwkCmd -f $LIB_DIR/db.selcc.awk $f.cc.txt > S_$$
diff --git a/detectors/cds/tools/chlorodb/subdb/lib/db.cc.awk b/detectors/cds/tools/chlorodb/subdb/lib/db.cc.awk
new file mode 100644
index 0000000..cb64a1e
--- /dev/null
+++ b/detectors/cds/tools/chlorodb/subdb/lib/db.cc.awk
@@ -0,0 +1,31 @@
+#
+#
+
+function visit(u, i, _local_, v) {
+  if (Visited[u]) return
+  Visited[u] = i
+  for (v in Edge[u]) {
+    visit(v, i)
+  }
+}
+
+/^#/ { next }
+
+{
+  Node[$1]++
+  Node[$2]++
+  Edge[$1][$2]++
+  Edge[$2][$1]++
+}
+
+END {
+
+  for (u in Node) {
+    if (Visited[u]) continue
+    visit(u, ++NbComp)
+  }
+
+  for (u in Node)
+    print u, Visited[u]
+  
+}
\ No newline at end of file
diff --git a/detectors/cds/tools/chlorodb/subdb/lib/db.cc.r b/detectors/cds/tools/chlorodb/subdb/lib/db.cc.r
deleted file mode 100755
index 2a69b8e..0000000
--- a/detectors/cds/tools/chlorodb/subdb/lib/db.cc.r
+++ /dev/null
@@ -1,18 +0,0 @@
-#!/usr/bin/env Rscript
-#
-
-require(igraph, warn.conflicts=F)
-
-args <- commandArgs(T)
-path <- if(length(args) > 0) args[1] else 'graph.dl'
-
-g <- read.graph(path, format='dl')
-
-cc <- clusters(g)
-
-res <- cbind(V(g)$name, membership(cc))
-
-write.table(res, quote=FALSE, row.names=FALSE, col.names=FALSE)
-
-quit(save="no")
-
diff --git a/detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.awk b/detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.awk
new file mode 100644
index 0000000..1e059a4
--- /dev/null
+++ b/detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.awk
@@ -0,0 +1,33 @@
+#
+
+function Abs(x) {
+  return (x+0 >= 0) ? x+0 : 0-x
+}
+
+function Median(array, _local_, tmp, size, lhs, rhs) {
+  size = asort(array, tmp)
+  lhs = int((size - 1) / 2) + 1
+  rhs = int(size / 2) + 1
+  return ((tmp[lhs] + tmp[rhs]) / 2.0)
+}
+
+#
+
+BEGIN {
+  if (DELTA=="") DELTA=0.5
+}
+
+/^>/ {
+  N++
+  Id[N] = substr($1, 2)
+  na = split($1, a, "@")
+  Len[N] = a[na]
+}
+
+END {
+  med = Median(Len)
+  for (i = 1 ; i <= N ; i++) {
+    if (Abs(Len[i]-med)/med <= DELTA)
+      print Id[i]
+  }
+}
diff --git a/detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.r b/detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.r
deleted file mode 100755
index 13cb5a2..0000000
--- a/detectors/cds/tools/chlorodb/subdb/lib/db.filter.len.r
+++ /dev/null
@@ -1,19 +0,0 @@
-#!/usr/bin/env Rscript
-#
-
-args <- commandArgs(T)
-path  <- if(length(args) > 0) args[1] else 'len.txt'
-delta <- if(length(args) > 1) args[2] else 0.5
-
-tab <- read.table(path, header=T)
-
-lmed <- median(tab$len)
-
-dlen <- lmed * as.numeric(delta)
-
-tab$ok <- (abs(tab$len-lmed)/lmed) <= delta
-
-write.table(tab, quote=F)
-
-quit(save='no')
-
diff --git a/detectors/cds/tools/chlorodb/subdb/lib/db.getlen.awk b/detectors/cds/tools/chlorodb/subdb/lib/db.getlen.awk
deleted file mode 100644
index 117c90b..0000000
--- a/detectors/cds/tools/chlorodb/subdb/lib/db.getlen.awk
+++ /dev/null
@@ -1,10 +0,0 @@
-#
-BEGIN {
-  print "id len"
-}
-
-/^>/ {
- na = split($1, a, "@")
- print substr($1, 2), a[na]
-}
-
diff --git a/detectors/cds/tools/lib/install.rpackages.r b/detectors/cds/tools/lib/install.rpackages.r
index c6f9778..9bc9175 100755
--- a/detectors/cds/tools/lib/install.rpackages.r
+++ b/detectors/cds/tools/lib/install.rpackages.r
@@ -25,7 +25,6 @@ check("grid")
 check("gridExtra")
 check("vcd")
 check("plotrix")
-check("igraph")
 
 quit(save='no', status=0)