Repository 'detrprok_scripts'
hg clone https://toolshed.g2.bx.psu.edu/repos/clairetn/detrprok_scripts

Changeset 0:a53eb951b164 (2013-03-25)
Next changeset 1:2d99d84922c7 (2013-04-26)
Commit message:
Uploaded
added:
detrprok_scripts/colorGff.pl
detrprok_scripts/colorGff.xml
detrprok_scripts/listGff.sh
detrprok_scripts/listGff.xml
detrprok_scripts/splitTranscriptGff.pl
detrprok_scripts/splitTranscriptGff.xml
detrprok_scripts/test-data/colorGff_Reference.gff
detrprok_scripts/test-data/colorGff_Result.gff
detrprok_scripts/test-data/listGff_Reference.gff
detrprok_scripts/test-data/listGff_Results.txt
detrprok_scripts/test-data/splitTranscript_Reference.gff
detrprok_scripts/test-data/splitTranscript_Result.gff
detrprok_scripts/test-data/splitTranscript_TranscriptionUnit.gff
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/colorGff.pl
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/colorGff.pl Mon Mar 25 05:39:29 2013 -0400
[
@@ -0,0 +1,69 @@
+#!/usr/bin/perl -w
+###
+# But : ajout ou modif de couleur d'un gff
+# 
+# Entrees : fichier gff
+#
+# Sortie : gff affiche a l'ecran
+#
+###------------------------------------------------------
+use vars qw($USAGE);                      
+use strict;                               
+
+=head1 NAME
+
+colorGff.pl - add or change color of a gff file
+
+=head1 SYNOPSIS
+
+% colorGff.pl -i file.gff -c color [-h] 
+
+=head1 DESCRIPTION
+This script will add or replace the 'color' tag of a gff formated file.
+
+    -i|--input fileName  gff input file name
+    -c|--color RGBcode   RGB code for color
+   [-h|--help]           help mode then die                              
+
+=head1 AUTHOR - Claire Toffano-Nioche - jan.11
+
+=cut
+#-----------------------
+my ($fileName, $colorGff) = ("", "") ;
+   # command line check
+    foreach my $num (0 .. $#ARGV) {
+        SWITCH: for ($ARGV[$num]) {
+        /--input|-i/ && do { 
+ $fileName=$ARGV[$num+1]; 
+ open ( fichierGff, "< $fileName" ) or die "Can't open gff file: \"$fileName\"\n" ; 
+ last };
+        /--color|-c/ && do {
+ $colorGff =$ARGV[$num+1]." ".$ARGV[$num+2]." ".$ARGV[$num+3];
+ last };
+        /--help|-h/ && do { exec("pod2text $0\n") ; die };
+        }
+    }
+
+    # informations retrieval
+    my @lines = <fichierGff> ; 
+    close fichierGff ;
+    # treatment
+ #print "gff file read ; number of lines : $#lines\n";
+    for (my $i=0 ; $i <= $#lines ; $i++) {
+ if (defined ($lines[$i])) {
+     if (!($lines[$i] =~ m/^\s*$|^#/)) { # skip both null and comment lines
+ if ($lines[$i] =~ /;/) {
+    if ($lines[$i] =~ /color=/) { # replace value
+      $lines[$i] =~ s/color=.*;*/color=$colorGff;/ ;
+    } else { # add color tag and value
+      $lines[$i] =~ s/;/;color=$colorGff;/ ;
+         } 
+         } else { # only one tag so no coma: add color tag
+         chomp($lines[$i]) ;
+         $lines[$i] .= "; color=".$colorGff.";\n";
+         }
+     }
+            print $lines[$i];
+        }
+    }
+exit(0);
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/colorGff.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/colorGff.xml Mon Mar 25 05:39:29 2013 -0400
[
@@ -0,0 +1,27 @@
+<tool id="colorGff" name="colorGff">
+ <description> Add or replace a "color" tag to a gff format file. This tag is use by Artemis (the genome browser and annotation tool of Sanger Institute). </description>
+ <command interpreter="perl"> 
+ colorGff.pl -i $referenciesFile -c $RGBcolor > $outputFile  
+ </command>
+
+ <inputs>
+ <param name="referenciesFile" type="data" label="Input File" format="gff" />
+ <param name="RGBcolor" type="text" value="None" label="RGB color" help="Give 3 integer values (from 0 to 255, space separated), each of them corresponding to the combination of the Red, Green and Blue color space. Ex.:'255 0 100'" />
+ </inputs>
+
+ <outputs>
+ <data format="gff" name="outputFile" label="[colorGff] Output File"/>
+ </outputs>
+
+        <tests>
+                <test>
+                        <param name="referenciesFile" value="colorGff_Reference.gff" />
+                        <param name="RGBcolor" value="255 0 100" />
+                        <output name="outputFile" file="colorGff_Result.gff" />
+                </test>
+        </tests>
+
+ <help>
+
+ </help>
+</tool>
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/listGff.sh
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/listGff.sh Mon Mar 25 05:39:29 2013 -0400
[
@@ -0,0 +1,3 @@
+#!/bin/bash
+awk '!/^\s*$|^#/{print $3}' $1 | grep "[[:alpha:]]" | sort -n | uniq -c
+
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/listGff.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/listGff.xml Mon Mar 25 05:39:29 2013 -0400
[
@@ -0,0 +1,21 @@
+<tool id="listGff" name="listGff">
+ <description> List and count features (column 3) of a gff file </description>
+ <command interpreter="sh"> 
+ listGff.sh $referenciesFile > $outputFile  
+ </command>
+
+ <inputs>
+ <param name="referenciesFile" type="data" label="Referencies Input File" format="gff" />
+ </inputs>
+
+ <outputs>
+ <data format="txt" name="outputFile" label="[listGff] Output File"/>
+ </outputs>
+
+        <tests>
+                <test>
+                        <param name="referenciesFile" value="listGff_Reference.gff" />
+                        <output name="outputFile" file="listGff_Results.txt" />
+                </test>
+        </tests>
+</tool>
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/splitTranscriptGff.pl
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/splitTranscriptGff.pl Mon Mar 25 05:39:29 2013 -0400
[
b'@@ -0,0 +1,197 @@\n+#!/usr/bin/perl -w\n+###\n+# But : croiser\n+# \n+# Entrees : 2 fichiers gff \xc3\xa0 croiser\n+#\n+# Sortie : gff affiche a l\'ecran\n+#\n+###------------------------------------------------------\n+use vars qw($USAGE);                      \n+use strict;                               \n+\n+=head1 NAME\n+\n+splitTranscriptGff.pl - compare 2 input gff files and define intervalls by couple of overlapping elements\n+\n+=head1 SYNOPSIS\n+\n+% intervallsExtractorGff.pl -i file1.gff -j file2.gff -s strand [-h] \n+\n+=head1 DESCRIPTION\n+This script will sort 2 gff files and compute distance between 2 successives lines.\n+\n+    -i|--input1 fileName   gff input file name: included elements\n+    -j|--input2 fileName   gff input file name: extended elements\n+   [-s|--strand] [s|d]\t   s for single strand (colinear) or d for double strands (antisense) [default d]\n+   [-h|--help]             help mode then die                              \n+\n+=head1 USECASE\n+Compare 2 input gff files: an annotations file (included elements) and transcription units file (extended elements).\n+For each couple of overlapping elements, split the transcription unit in 5\'utr, CDS, 3\'utr or "operon" in case of successives genes included in the transcription unit.\n+\n+=head1 KWON BUG\n+Fisrt and last elements on the genome sequence can be misclassified in the result file.\n+\n+=head1 AUTHOR\n+Claire Toffano-Nioche - sep.11\n+\n+=cut\n+#-----------------------\n+sub feedPositionTab { my ($val, $pF, $pB, @info) = @_ ;\n+\t\t#print "feedPositionTab::$#info, ", ($#info+1)/4," \\n";\n+\tfor (my $i=0 ; $i <= $#info ; $i+=4) { # for each extended element \n+\t\t\t#print "....$info[$i+2]\\n";\n+\t\tif ($info[$i+3] =~ /\\+/) {\n+\t\t\tfor (my $c = $info[$i+1] ; $c <= $info[$i+2] ; $c++) { @$pF[$c]=$val } ; # Forward sequence\n+\t\t} else {\n+\t\t\tfor (my $c = $info[$i+1] ; $c <= $info[$i+2] ; $c++) { @$pB[$c]=$val } ; # Backward sequence\n+\t\t}\n+\t}\n+\t\t#print "feedPos...:: ", join(".", @$pF[0..100]), "\\n";\n+\t\t#print "feedPos...:: ", join(".", @$pB[0..100]), "\\n";\n+}\n+#-----------------------\n+sub recupInfo {\tmy ($pInfo, @lines) = @_ ;\n+    my $i=0 ;\n+    while ($i<=$#lines){\n+\tchomp($lines[$i]);\n+\tmy @line = split("\\t",$lines[$i]);\n+\tif (defined ($line[0])) {\n+\t\tif (!($line[0] =~ m/^\\s*$|^#/)) { # skip both null and comment lines\n+\t\t\tpush(@$pInfo, $line[0], $line[3], $line[4], $line[6]) ; # 0=name, 3=begin, 4=end, 6=strand\n+\t\t} \n+\t}\n+\t$i=$i+1 ;\n+    }\n+    \t#print "recupInfo::end=", $i, "\\n" ;\n+}\n+#-----------------------\n+sub tagName { my ($seqN, $posB, $posE, $strand) = @_ ;    \n+\tmy $tagN=$seqN.$strand.$posB."..".$posE;\n+\t\t#print "tagName::",join("_",@_)," and tagName:$tagN\\n";\n+return $tagN;\n+}\n+#-----------------------\n+sub transitionAnalysis {\n+my ($pos, $seq, $s, $pdebAmont, $pfinAmont, $pdebIn, $pfinIn, $pdebAval, $pfinAval, $ptag) = @_ ;\n+    my $enCours = @$ptag[$pos] ; \n+    my $precedant = ($s =~ /\\+/?@$ptag[$pos-1]:@$ptag[$pos+1]) ; \n+    if ($enCours ne $precedant) {\n+    \t#print "transi...:: $s, $pos, $precedant, $enCours\\n";\n+    \t#print "transition::$$pdebAmont, $$pfinAmont, $$pdebIn, $$pfinIn, $$pdebAval, $$pfinAval\\n";\n+    \tSWITCH: for ($precedant.$enCours) {\n+               \t/01/ && do { $$pdebAmont = $pos ; last SWITCH ;};\n+                /02/ && do { $$pdebIn = $pos ; last SWITCH ;};\n+                /10/ && do { $$pfinAval = ($s =~/\\+/?$pos-1:$pos+1) ; \n+                \t\tif (($s =~ /\\+/)and ($$pdebAval!=$$pfinAval)) {\n+                \t\t\tprintf "%s\\tsplit\\tutr3\\t%s\\t%s\\t.\\t%s\\t.\\tName=%s;\\n", \n+                \t\t\t\t$seq, $$pdebAval, $$pfinAval, $s, &tagName($seq, $$pdebAval, $$pfinAval, $s) ; \n+                \t\t\t#if ($$pdebAval==$$pfinAval) { print "transition 10 +\\n"};\n+                \t\t} elsif ($$pfinAval!=$$pdebAval) {\n+                \t\t\tprintf "%s\\tsplit\\tutr3\\t%s\\t%s\\t.\\t%s\\t.\\tName=%s;\\n", \n+                \t\t\t\t$seq, $$pfinAval, $$pdebAval, $s, &tagName($seq, $$pfinAval, $$pdebAval, $s) ; \n+                \t\t\t#if ($$pfinAval==$$pdebAval){ print "transition 10 -\\n"};\n+                \t'..b' \t\tlast SWITCH ;\n+                \t };\n+                /20/ && do { $$pfinIn=($s =~/\\+/?$pos-1:$pos+1) ; \n+                        if (($s =~ /\\+/) and ($$pdebIn!=$$pfinIn)) {\n+                        \tprintf "%s\\tsplit\\tgene\\t%s\\t%s\\t.\\t%s\\t.\\tName=%s;\\n", \n+                \t\t\t\t$seq, $$pdebIn, $$pfinIn, $s, &tagName($seq, $$pdebIn, $$pfinIn, $s) ; \n+                \t\t} elsif ($$pfinIn!=$$pdebIn) {\n+                \t\t    printf "%s\\tsplit\\tgene\\t%s\\t%s\\t.\\t%s\\t.\\tName=%s;\\n", \n+                \t\t\t\t$seq, $$pfinIn, $$pdebIn, $s, &tagName($seq, $$pfinIn, $$pdebIn, $s) ; \n+                \t\t}\n+                \t\t$$pdebIn = 0 ; $$pfinIn = 0 ;\n+                \t\tlast SWITCH ;\n+                \t };\n+                /21/ && do { $$pdebAval=$pos ; $$pfinIn=($s =~/\\+/?$pos-1:$pos+1) ; \n+                        if (($s =~ /\\+/) and ($$pdebIn!=$$pfinIn)) {\n+                        \tprintf "%s\\tsplit\\tgene\\t%s\\t%s\\t.\\t%s\\t.\\tName=%s;\\n", \n+                \t\t\t\t$seq, $$pdebIn, $$pfinIn, $s, &tagName($seq, $$pdebIn, $$pfinIn, $s) ; \n+                \t\t} elsif ($$pfinIn!=$$pdebIn) {\n+                \t\t\tprintf "%s\\tsplit\\tgene\\t%s\\t%s\\t.\\t%s\\t.\\tName=%s;\\n", \n+                \t\t\t\t$seq, $$pfinIn, $$pdebIn, $s, &tagName($seq, $$pfinIn, $$pdebIn, $s) ; \n+                \t\t}\n+                \t\t#$$pdebIn = 0 ; $$pfinIn = 0 ;\n+                \t\tlast SWITCH ;\n+                \t };\n+          }\n+    }\n+ }\n+#-----------------------\t\n+my ($fileNameI, $fileNameE, $strand) = ("", "", 0) ;\n+# command line check\n+foreach my $num (0 .. $#ARGV) {\n+        SWITCH: for ($ARGV[$num]) {\n+        /--input1|-i/ && do { \n+\t\t\t$fileNameI=$ARGV[$num+1]; \n+\t\t\topen ( fichierGffI, "< $fileNameI" ) or die "Can\'t open gff file: \\"$fileNameI\\"\\n" ; \n+\t\t\tlast };\n+\t/--input2|-j/ && do { \n+\t\t\t$fileNameE=$ARGV[$num+1]; \n+\t\t\topen ( fichierGffE, "< $fileNameE" ) or die "Can\'t open gff file: \\"$fileNameE\\"\\n" ; \n+\t\t\tlast };\n+        /--strand|-s/ && do { \n+\t\t\tif ($ARGV[$num+1] eq "s") { $strand=1}; \n+\t\t\tlast };\n+        /--help|-h/ && do { exec("pod2text $0\\n") ; die };\n+        }\n+}\n+# memory declarations:\n+my @infoI ; my @infoE ;\n+my $seqName ;\n+my @tagF ; my @tagB ; # for Forward and Backward sequences\n+# data retrieval:\n+my @linesI = <fichierGffI> ; my @linesE = <fichierGffE> ;\n+close fichierGffI ; close fichierGffE ;\n+\t#print "gff files read ; number of lines : ",$#linesI+1," + ",$#linesE+1,"\\n";\n+# positions management:\n+&recupInfo(\\@infoI, @linesI) ; \n+\t#print "number of treated elements:",($#infoI+1)/4,"\\n";\n+&recupInfo(\\@infoE, @linesE) ;\n+\t#print "number of treated elements:",($#infoE+1)/4,"\\n";\n+# treatement: \n+# transform gff lines into chromosomal position tags : 0 for nothing, 1 resp. 2 for extended resp. included elements\n+if (($#infoI) and ($#infoE)) { \n+\t$seqName=$infoI[0] ;\n+\t\t#print "end : $infoE[$#infoE-1]\\n";\n+\tfor (my $i=0 ; $i <= $infoE[$#infoE-1] ; $i++) { $tagF[$i] = 0 ; $tagB[$i] = 0 ; } ; # "O" tag in all chr. positions\n+\t\t#print "seqName : $seqName\\n" ;\n+\t&feedPositionTab(1, \\@tagF, \\@tagB, @infoE) ; # "1" tag for all extended elements\n+\t&feedPositionTab(2, \\@tagF, \\@tagB, @infoI) ; # "2" tag for all included elements\n+\t\t#print join("", @tagF), "\\n";\n+\t\t#print join("", @tagB), "\\n";\n+\t# transition management:\n+\tmy ($beginUpstream, $endUpstream, $beginIncluded, $endIncluded, $beginDownstream, $endDownstream) \n+\t\t= (0, 0, 0, 0, 0, 0) ;\n+\tfor (my $i=1 ; $i <= $#tagF-1 ; $i+=1) {\n+\t\t&transitionAnalysis($i, $seqName, "+", \\$beginUpstream, \\$endUpstream, \\$beginIncluded, \\$endIncluded, \\$beginDownstream, \\$endDownstream, \\@tagF) ;\n+\t}\n+\t($beginUpstream, $endUpstream, $beginIncluded, $endIncluded, $beginDownstream, $endDownstream) \n+\t\t= ($infoE[$#infoE-1], $infoE[$#infoE-1], $infoE[$#infoE-1], $infoE[$#infoE-1], $infoE[$#infoE-1], $infoE[$#infoE-1]) ;\n+\tfor (my $i=$#tagB-1 ; $i >= 1 ; $i-=1) {\n+\t\t&transitionAnalysis($i, $seqName, "-", \\$beginUpstream, \\$endUpstream, \\$beginIncluded, \\$endIncluded, \\$beginDownstream, \\$endDownstream, \\@tagB) ;\n+\t}\n+}\n+exit(0) ;\n'
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/splitTranscriptGff.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/splitTranscriptGff.xml Mon Mar 25 05:39:29 2013 -0400
[
@@ -0,0 +1,27 @@
+<tool id="splitTranscriptGff" name="splitTranscriptGff">
+ <description> Define UTRs and intergenic operonic regions from a transcript file and following a reference file </description>
+ <command interpreter="perl"> 
+ splitTranscriptGff.pl -i $referenciesFile -j $transcriptsFile > $outputFile  
+ </command>
+
+ <inputs>
+ <param name="referenciesFile" type="data" label="Referencies Input File" format="gff" />
+ <param name="transcriptsFile" type="data" label="Transcripts Input File" format="gff" />
+ </inputs>
+
+ <outputs>
+ <data format="gff3" name="outputFile" label="[splitTranscript] Output File"/>
+ </outputs>
+
+        <tests>
+                <test>
+                        <param name="referenciesFile" value="splitTranscript_Reference.gff" />
+                        <param name="transcriptsFile" value="splitTranscript_TranscriptionUnit.gff" />
+                        <output name="outputFile" file="splitTranscript_Result.gff" />
+                </test>
+        </tests>
+
+ <help>
+ Note that inputs files should be sorted by increasing positions and that referencies should be included in transcripts.
+ </help>
+</tool>
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/colorGff_Reference.gff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/colorGff_Reference.gff Mon Mar 25 05:39:29 2013 -0400
b
@@ -0,0 +1,10 @@
+##gff-version 3
+#!gff-spec-version 1.20
+#!processor NCBI annotwriter
+##sequence-region NC_007795.1 1 28213
+##species http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=93061
+NC_007795.1 RefSeq region 1 28213 . + . ID=id0;Dbxref=taxon:93061;Is_circular=true;gbkey=Src;genome=chromosome;mol_type=genomic DNA;strain=NCTC 8325;sub-species=aureus
+NC_007795.1 RefSeq gene 517 1878 . + . ID=gene0;Name=dnaA;Dbxref=GeneID:3919798;gbkey=Gene;gene=dnaA;locus_tag=SAOUHSC_00001;color=127 255 312
+NC_007795.1 RefSeq CDS 517 1878 . + 0 ID=cds0;Name=YP_498609.1;Parent=gene0;Note=binds to the dnaA-box as an ATP-bound complex at the origin of replication during the initiation of chromosomal replication%3B can also affect transcription of multiple genes including itself.;Dbxref=Genbank:YP_498609.1,GeneID:3919798;gbkey=CDS;product=chromosomal replication initiation protein;protein_id=YP_498609.1;transl_table=11
+NC_007795.1 RefSeq gene 2156 3289 . + . ID=gene1;Name=SAOUHSC_00002;Dbxref=GeneID:3919799;gbkey=Gene;locus_tag=SAOUHSC_00002;color=127 255 312;
+NC_007795.1 RefSeq CDS 2156 3289 . + 0 ID=cds1;Name=YP_498610.1;Parent=gene1;Note=binds the polymerase to DNA and acts as a sliding clamp;Dbxref=Genbank:YP_498610.1,GeneID:3919799;gbkey=CDS;color=127 255 312;product=DNA polymerase III subunit beta;protein_id=YP_498610.1;transl_table=11
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/colorGff_Result.gff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/colorGff_Result.gff Mon Mar 25 05:39:29 2013 -0400
b
@@ -0,0 +1,10 @@
+##gff-version 3
+#!gff-spec-version 1.20
+#!processor NCBI annotwriter
+##sequence-region NC_007795.1 1 28213
+##species http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=93061
+NC_007795.1 RefSeq region 1 28213 . + . ID=id0;color=255 0 100;Dbxref=taxon:93061;Is_circular=true;gbkey=Src;genome=chromosome;mol_type=genomic DNA;strain=NCTC 8325;sub-species=aureus
+NC_007795.1 RefSeq gene 517 1878 . + . ID=gene0;Name=dnaA;Dbxref=GeneID:3919798;gbkey=Gene;gene=dnaA;locus_tag=SAOUHSC_00001;color=255 0 100;
+NC_007795.1 RefSeq CDS 517 1878 . + 0 ID=cds0;color=255 0 100;Name=YP_498609.1;Parent=gene0;Note=binds to the dnaA-box as an ATP-bound complex at the origin of replication during the initiation of chromosomal replication%3B can also affect transcription of multiple genes including itself.;Dbxref=Genbank:YP_498609.1,GeneID:3919798;gbkey=CDS;product=chromosomal replication initiation protein;protein_id=YP_498609.1;transl_table=11
+NC_007795.1 RefSeq gene 2156 3289 . + . ID=gene1;Name=SAOUHSC_00002;Dbxref=GeneID:3919799;gbkey=Gene;locus_tag=SAOUHSC_00002;color=255 0 100;
+NC_007795.1 RefSeq CDS 2156 3289 . + 0 ID=cds1;Name=YP_498610.1;Parent=gene1;Note=binds the polymerase to DNA and acts as a sliding clamp;Dbxref=Genbank:YP_498610.1,GeneID:3919799;gbkey=CDS;color=255 0 100;
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/listGff_Reference.gff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/listGff_Reference.gff Mon Mar 25 05:39:29 2013 -0400
b
b'@@ -0,0 +1,1581 @@\n+NC_011744\tS-MART\tCDS\t409\t4831\t.\t+\t.\tproduct=hypothetical protein;nbElements=3.000000;transl_table=11;gbkey=CDS;protein_id=YP_002394608.1;Dbxref=Genbank:YP_002394608.1,GeneID:7137318;ID=cds2;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1\n+NC_011744\tS-MART\texon\t409\t1626\t.\t+\t.\tID=cds2-exon1;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1-exon1;Parent=cds2\n+NC_011744\tS-MART\texon\t1632\t2606\t.\t+\t.\tID=cds2-exon2;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1-exon2;Parent=cds2\n+NC_011744\tS-MART\texon\t2699\t4831\t.\t+\t.\tID=cds2-exon3;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1-exon3;Parent=cds2\n+NC_011744\tS-MART\tCDS\t5362\t7032\t.\t+\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394609.1;Dbxref=Genbank:YP_002394609.1,GeneID:7137319;ID=cds3;Name=YP_002394609.1\n+NC_011744\tS-MART\tCDS\t7180\t16787\t.\t-\t.\tproduct=Signal transduction histidine kinase regulating C4-dicarboxylate transport system;nbElements=6.000000;transl_table=11;gbkey=CDS;protein_id=YP_002394615.1;Dbxref=Genbank:YP_002394615.1,GeneID:7137325;ID=cds9;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1\n+NC_011744\tS-MART\texon\t7180\t9294\t.\t-\t.\tID=cds9-exon1;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon1;Parent=cds9\n+NC_011744\tS-MART\texon\t9312\t11222\t.\t-\t.\tID=cds9-exon2;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon2;Parent=cds9\n+NC_011744\tS-MART\texon\t11228\t11896\t.\t-\t.\tID=cds9-exon3;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon3;Parent=cds9\n+NC_011744\tS-MART\texon\t11956\t13347\t.\t-\t.\tID=cds9-exon4;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon4;Parent=cds9\n+NC_011744\tS-MART\texon\t13460\t14773\t.\t-\t.\tID=cds9-exon5;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon5;Parent=cds9\n+NC_011744\tS-MART\texon\t14865\t16787\t.\t-\t.\tID=cds9-exon6;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon6;Parent=cds9\n+NC_011744\tS-MART\tCDS\t16954\t17589\t.\t+\t.\tproduct=pyridoxamine 5%27-phosphate oxidase;transl_table=11;Note=catalyzes the formation of pyridoxal 5%27-phosphate from pyridoxamine 5%27-phosphate;gbkey=CDS;protein_id=YP_002394616.1;Dbxref=Genbank:YP_002394616.1,GeneID:7137326;ID=cds10;Name=YP_002394616.1\n+NC_011744\tS-MART\tCDS\t17694\t18476\t.\t-\t.\tproduct=LuxR family transcriptional regulator;transl_table=11;gbkey=CDS;protein_id=YP_002394617.1;Dbxref=Genbank:YP_002394617.1,GeneID:7137327;ID=cds11;Name=YP_002394617.1\n+NC_011744\tS-MART\tCDS\t18853\t20682\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394618.1;Dbxref=Genbank:YP_002394618.1,GeneID:7137328;ID=cds12;Name=YP_002394618.1\n+NC_011744\tS-MART\tCDS\t20840\t21961\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394619.1;Dbxref=Genbank:YP_002394619.1,GeneID:7137329;ID=cds13;Name=YP_002394619.1\n+NC_011744\tS-MART\tCDS\t22321\t22524\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394620.1;Dbxref=Genbank:YP_002394620.1,GeneID:7137330;ID=cds14;Name=YP_002394620.1\n+NC_011744\tS-MART\tCDS\t22448\t23272\t.\t+\t.\tproduct=AraC family transcriptional regulator;transl_table=11;gbkey=CDS;protein_id=YP_002394621.1;Dbxref=Genbank:YP_002394621.1,GeneID:7137331;ID=cds15;Name=YP_002394621.1\n+NC_011744\tS-MART\tCDS\t23454\t27325\t.\t+\t.\tproduct=hypothetical protein;nbElements=2.000000;transl_table=11;gbkey=CDS;protein_id=YP_002394623.1;Dbxref=Genbank:YP_002394623.1,GeneID:7137333;ID=cds17;Name=YP_002394623.1--YP_002394622.1\n+NC_011744\tS-MART\texon\t23454\t26609\t.\t+\t.\tID=cds17-exon1;Name=YP_002394623.1--YP_002394622.1-exon1;Parent=cds17\n+NC_011744\tS-MART\texon\t26621\t27325\t.\t+\t.\tID=cds17-exon2;Name=YP_002394623.1--YP_002394622.1-exon2;Parent=cds17\n+NC_011744\tS-MART\tCDS\t2750'..b'468-exon2;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon2;Parent=cds1468\n+NC_011744\tS-MART\texon\t1654998\t1656358\t.\t+\t.\tID=cds1468-exon3;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon3;Parent=cds1468\n+NC_011744\tS-MART\texon\t1656418\t1657038\t.\t+\t.\tID=cds1468-exon4;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon4;Parent=cds1468\n+NC_011744\tS-MART\texon\t1657098\t1657718\t.\t+\t.\tID=cds1468-exon5;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon5;Parent=cds1468\n+NC_011744\tS-MART\tCDS\t1658632\t1659561\t.\t+\t.\tproduct=carbamate kinase;transl_table=11;Note=reversible synthesis of carbamate and ATP from carbamoyl phosphate and ADP;gbkey=CDS;protein_id=YP_002396075.1;Dbxref=Genbank:YP_002396075.1,GeneID:7138782;ID=cds1469;Name=YP_002396075.1\n+NC_011744\tS-MART\tCDS\t1659797\t1663769\t.\t-\t.\tproduct=4-hydroxyphenylpyruvate dioxygenase;nbElements=4.000000;transl_table=11;Note=Evidence 2a : Function of homologous gene experimentally demonstrated in another organism%3B PubMedId : 8112844%3B Product type e : enzyme;gbkey=CDS;protein_id=YP_002396079.1;Dbxref=Genbank:YP_002396079.1,GeneID:7138786;ID=cds1473;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1\n+NC_011744\tS-MART\texon\t1659797\t1660501\t.\t-\t.\tID=cds1473-exon1;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1-exon1;Parent=cds1473\n+NC_011744\tS-MART\texon\t1660523\t1661560\t.\t-\t.\tID=cds1473-exon2;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1-exon2;Parent=cds1473\n+NC_011744\tS-MART\texon\t1661573\t1663769\t.\t-\t.\tID=cds1473-exon3;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1-exon3;Parent=cds1473\n+NC_011744\tS-MART\tCDS\t1663853\t1664035\t.\t+\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396080.1;Dbxref=Genbank:YP_002396080.1,GeneID:7138787;ID=cds1474;Name=YP_002396080.1\n+NC_011744\tS-MART\tCDS\t1664130\t1664789\t.\t-\t.\tproduct=phosphoserine phosphatase;transl_table=11;gbkey=CDS;protein_id=YP_002396081.1;Dbxref=Genbank:YP_002396081.1,GeneID:7138788;ID=cds1475;Name=YP_002396081.1\n+NC_011744\tS-MART\tCDS\t1665142\t1667568\t.\t-\t.\tproduct=two-component sensor;transl_table=11;gbkey=CDS;protein_id=YP_002396082.1;Dbxref=Genbank:YP_002396082.1,GeneID:7138789;ID=cds1476;Name=YP_002396082.1\n+NC_011744\tS-MART\tCDS\t1667657\t1668415\t.\t+\t.\tproduct=amino acid ABC transporter substrate-binding protein;transl_table=11;gbkey=CDS;protein_id=YP_002396083.1;Dbxref=Genbank:YP_002396083.1,GeneID:7138790;ID=cds1477;Name=YP_002396083.1\n+NC_011744\tS-MART\tCDS\t1668625\t1669185\t.\t-\t.\tproduct=GTP cyclohydrolase II;transl_table=11;gbkey=CDS;protein_id=YP_002396084.1;Dbxref=Genbank:YP_002396084.1,GeneID:7138791;ID=cds1478;Name=YP_002396084.1\n+NC_011744\tS-MART\tCDS\t1669376\t1669699\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396085.1;Dbxref=Genbank:YP_002396085.1,GeneID:7138792;ID=cds1479;Name=YP_002396085.1\n+NC_011744\tS-MART\tCDS\t1669923\t1672240\t.\t+\t.\tproduct=hypothetical protein;nbElements=3.000000;transl_table=11;gbkey=CDS;protein_id=YP_002396088.1;Dbxref=Genbank:YP_002396088.1,GeneID:7138795;ID=cds1482;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1\n+NC_011744\tS-MART\texon\t1669923\t1670756\t.\t+\t.\tID=cds1482-exon1;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1-exon1;Parent=cds1482\n+NC_011744\tS-MART\texon\t1670766\t1672240\t.\t+\t.\tID=cds1482-exon2;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1-exon2;Parent=cds1482\n+NC_011744\tS-MART\tCDS\t1672182\t1672271\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396089.1;Dbxref=Genbank:YP_002396089.1,GeneID:7138796;ID=cds1483;Name=YP_002396089.1\n+NC_011744\tS-MART\tCDS\t1672683\t1674662\t.\t-\t.\tproduct=RctB protein;transl_table=11;gbkey=CDS;protein_id=YP_002396090.1;Dbxref=Genbank:YP_002396090.1,GeneID:7138797;ID=cds1484;Name=YP_002396090.1\n'
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/listGff_Results.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/listGff_Results.txt Mon Mar 25 05:39:29 2013 -0400
b
@@ -0,0 +1,5 @@
+    871 CDS
+    699 exon
+      6 ncRNA
+      2 rRNA
+      3 tRNA
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/splitTranscript_Reference.gff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/splitTranscript_Reference.gff Mon Mar 25 05:39:29 2013 -0400
b
b'@@ -0,0 +1,1581 @@\n+NC_011744\tS-MART\tCDS\t409\t4831\t.\t+\t.\tproduct=hypothetical protein;nbElements=3.000000;transl_table=11;gbkey=CDS;protein_id=YP_002394608.1;Dbxref=Genbank:YP_002394608.1,GeneID:7137318;ID=cds2;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1\n+NC_011744\tS-MART\texon\t409\t1626\t.\t+\t.\tID=cds2-exon1;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1-exon1;Parent=cds2\n+NC_011744\tS-MART\texon\t1632\t2606\t.\t+\t.\tID=cds2-exon2;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1-exon2;Parent=cds2\n+NC_011744\tS-MART\texon\t2699\t4831\t.\t+\t.\tID=cds2-exon3;Name=YP_002394608.1--YP_002394607.1--YP_002394606.1-exon3;Parent=cds2\n+NC_011744\tS-MART\tCDS\t5362\t7032\t.\t+\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394609.1;Dbxref=Genbank:YP_002394609.1,GeneID:7137319;ID=cds3;Name=YP_002394609.1\n+NC_011744\tS-MART\tCDS\t7180\t16787\t.\t-\t.\tproduct=Signal transduction histidine kinase regulating C4-dicarboxylate transport system;nbElements=6.000000;transl_table=11;gbkey=CDS;protein_id=YP_002394615.1;Dbxref=Genbank:YP_002394615.1,GeneID:7137325;ID=cds9;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1\n+NC_011744\tS-MART\texon\t7180\t9294\t.\t-\t.\tID=cds9-exon1;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon1;Parent=cds9\n+NC_011744\tS-MART\texon\t9312\t11222\t.\t-\t.\tID=cds9-exon2;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon2;Parent=cds9\n+NC_011744\tS-MART\texon\t11228\t11896\t.\t-\t.\tID=cds9-exon3;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon3;Parent=cds9\n+NC_011744\tS-MART\texon\t11956\t13347\t.\t-\t.\tID=cds9-exon4;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon4;Parent=cds9\n+NC_011744\tS-MART\texon\t13460\t14773\t.\t-\t.\tID=cds9-exon5;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon5;Parent=cds9\n+NC_011744\tS-MART\texon\t14865\t16787\t.\t-\t.\tID=cds9-exon6;Name=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1-exon6;Parent=cds9\n+NC_011744\tS-MART\tCDS\t16954\t17589\t.\t+\t.\tproduct=pyridoxamine 5%27-phosphate oxidase;transl_table=11;Note=catalyzes the formation of pyridoxal 5%27-phosphate from pyridoxamine 5%27-phosphate;gbkey=CDS;protein_id=YP_002394616.1;Dbxref=Genbank:YP_002394616.1,GeneID:7137326;ID=cds10;Name=YP_002394616.1\n+NC_011744\tS-MART\tCDS\t17694\t18476\t.\t-\t.\tproduct=LuxR family transcriptional regulator;transl_table=11;gbkey=CDS;protein_id=YP_002394617.1;Dbxref=Genbank:YP_002394617.1,GeneID:7137327;ID=cds11;Name=YP_002394617.1\n+NC_011744\tS-MART\tCDS\t18853\t20682\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394618.1;Dbxref=Genbank:YP_002394618.1,GeneID:7137328;ID=cds12;Name=YP_002394618.1\n+NC_011744\tS-MART\tCDS\t20840\t21961\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394619.1;Dbxref=Genbank:YP_002394619.1,GeneID:7137329;ID=cds13;Name=YP_002394619.1\n+NC_011744\tS-MART\tCDS\t22321\t22524\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002394620.1;Dbxref=Genbank:YP_002394620.1,GeneID:7137330;ID=cds14;Name=YP_002394620.1\n+NC_011744\tS-MART\tCDS\t22448\t23272\t.\t+\t.\tproduct=AraC family transcriptional regulator;transl_table=11;gbkey=CDS;protein_id=YP_002394621.1;Dbxref=Genbank:YP_002394621.1,GeneID:7137331;ID=cds15;Name=YP_002394621.1\n+NC_011744\tS-MART\tCDS\t23454\t27325\t.\t+\t.\tproduct=hypothetical protein;nbElements=2.000000;transl_table=11;gbkey=CDS;protein_id=YP_002394623.1;Dbxref=Genbank:YP_002394623.1,GeneID:7137333;ID=cds17;Name=YP_002394623.1--YP_002394622.1\n+NC_011744\tS-MART\texon\t23454\t26609\t.\t+\t.\tID=cds17-exon1;Name=YP_002394623.1--YP_002394622.1-exon1;Parent=cds17\n+NC_011744\tS-MART\texon\t26621\t27325\t.\t+\t.\tID=cds17-exon2;Name=YP_002394623.1--YP_002394622.1-exon2;Parent=cds17\n+NC_011744\tS-MART\tCDS\t2750'..b'468-exon2;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon2;Parent=cds1468\n+NC_011744\tS-MART\texon\t1654998\t1656358\t.\t+\t.\tID=cds1468-exon3;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon3;Parent=cds1468\n+NC_011744\tS-MART\texon\t1656418\t1657038\t.\t+\t.\tID=cds1468-exon4;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon4;Parent=cds1468\n+NC_011744\tS-MART\texon\t1657098\t1657718\t.\t+\t.\tID=cds1468-exon5;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1-exon5;Parent=cds1468\n+NC_011744\tS-MART\tCDS\t1658632\t1659561\t.\t+\t.\tproduct=carbamate kinase;transl_table=11;Note=reversible synthesis of carbamate and ATP from carbamoyl phosphate and ADP;gbkey=CDS;protein_id=YP_002396075.1;Dbxref=Genbank:YP_002396075.1,GeneID:7138782;ID=cds1469;Name=YP_002396075.1\n+NC_011744\tS-MART\tCDS\t1659797\t1663769\t.\t-\t.\tproduct=4-hydroxyphenylpyruvate dioxygenase;nbElements=4.000000;transl_table=11;Note=Evidence 2a : Function of homologous gene experimentally demonstrated in another organism%3B PubMedId : 8112844%3B Product type e : enzyme;gbkey=CDS;protein_id=YP_002396079.1;Dbxref=Genbank:YP_002396079.1,GeneID:7138786;ID=cds1473;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1\n+NC_011744\tS-MART\texon\t1659797\t1660501\t.\t-\t.\tID=cds1473-exon1;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1-exon1;Parent=cds1473\n+NC_011744\tS-MART\texon\t1660523\t1661560\t.\t-\t.\tID=cds1473-exon2;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1-exon2;Parent=cds1473\n+NC_011744\tS-MART\texon\t1661573\t1663769\t.\t-\t.\tID=cds1473-exon3;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1-exon3;Parent=cds1473\n+NC_011744\tS-MART\tCDS\t1663853\t1664035\t.\t+\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396080.1;Dbxref=Genbank:YP_002396080.1,GeneID:7138787;ID=cds1474;Name=YP_002396080.1\n+NC_011744\tS-MART\tCDS\t1664130\t1664789\t.\t-\t.\tproduct=phosphoserine phosphatase;transl_table=11;gbkey=CDS;protein_id=YP_002396081.1;Dbxref=Genbank:YP_002396081.1,GeneID:7138788;ID=cds1475;Name=YP_002396081.1\n+NC_011744\tS-MART\tCDS\t1665142\t1667568\t.\t-\t.\tproduct=two-component sensor;transl_table=11;gbkey=CDS;protein_id=YP_002396082.1;Dbxref=Genbank:YP_002396082.1,GeneID:7138789;ID=cds1476;Name=YP_002396082.1\n+NC_011744\tS-MART\tCDS\t1667657\t1668415\t.\t+\t.\tproduct=amino acid ABC transporter substrate-binding protein;transl_table=11;gbkey=CDS;protein_id=YP_002396083.1;Dbxref=Genbank:YP_002396083.1,GeneID:7138790;ID=cds1477;Name=YP_002396083.1\n+NC_011744\tS-MART\tCDS\t1668625\t1669185\t.\t-\t.\tproduct=GTP cyclohydrolase II;transl_table=11;gbkey=CDS;protein_id=YP_002396084.1;Dbxref=Genbank:YP_002396084.1,GeneID:7138791;ID=cds1478;Name=YP_002396084.1\n+NC_011744\tS-MART\tCDS\t1669376\t1669699\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396085.1;Dbxref=Genbank:YP_002396085.1,GeneID:7138792;ID=cds1479;Name=YP_002396085.1\n+NC_011744\tS-MART\tCDS\t1669923\t1672240\t.\t+\t.\tproduct=hypothetical protein;nbElements=3.000000;transl_table=11;gbkey=CDS;protein_id=YP_002396088.1;Dbxref=Genbank:YP_002396088.1,GeneID:7138795;ID=cds1482;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1\n+NC_011744\tS-MART\texon\t1669923\t1670756\t.\t+\t.\tID=cds1482-exon1;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1-exon1;Parent=cds1482\n+NC_011744\tS-MART\texon\t1670766\t1672240\t.\t+\t.\tID=cds1482-exon2;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1-exon2;Parent=cds1482\n+NC_011744\tS-MART\tCDS\t1672182\t1672271\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396089.1;Dbxref=Genbank:YP_002396089.1,GeneID:7138796;ID=cds1483;Name=YP_002396089.1\n+NC_011744\tS-MART\tCDS\t1672683\t1674662\t.\t-\t.\tproduct=RctB protein;transl_table=11;gbkey=CDS;protein_id=YP_002396090.1;Dbxref=Genbank:YP_002396090.1,GeneID:7138797;ID=cds1484;Name=YP_002396090.1\n'
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/splitTranscript_Result.gff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/splitTranscript_Result.gff Mon Mar 25 05:39:29 2013 -0400
b
b'@@ -0,0 +1,911 @@\n+NC_011744\tsplit\tutr5\t364\t408\t.\t+\t.\tName=NC_011744+364..408;\n+NC_011744\tsplit\tgene\t409\t4831\t.\t+\t.\tName=NC_011744+409..4831;\n+NC_011744\tsplit\tutr5\t5319\t5361\t.\t+\t.\tName=NC_011744+5319..5361;\n+NC_011744\tsplit\tgene\t5362\t7032\t.\t+\t.\tName=NC_011744+5362..7032;\n+NC_011744\tsplit\tutr5\t16915\t16953\t.\t+\t.\tName=NC_011744+16915..16953;\n+NC_011744\tsplit\tgene\t16954\t17589\t.\t+\t.\tName=NC_011744+16954..17589;\n+NC_011744\tsplit\tutr3\t17590\t17644\t.\t+\t.\tName=NC_011744+17590..17644;\n+NC_011744\tsplit\tutr5\t22423\t22447\t.\t+\t.\tName=NC_011744+22423..22447;\n+NC_011744\tsplit\tgene\t22448\t23272\t.\t+\t.\tName=NC_011744+22448..23272;\n+NC_011744\tsplit\tutr5\t23395\t23453\t.\t+\t.\tName=NC_011744+23395..23453;\n+NC_011744\tsplit\tgene\t23454\t27325\t.\t+\t.\tName=NC_011744+23454..27325;\n+NC_011744\tsplit\toperon\t27326\t27508\t.\t+\t.\tName=NC_011744+27326..27508;\n+NC_011744\tsplit\tgene\t27509\t29002\t.\t+\t.\tName=NC_011744+27509..29002;\n+NC_011744\tsplit\tutr3\t29003\t29212\t.\t+\t.\tName=NC_011744+29003..29212;\n+NC_011744\tsplit\tutr5\t29339\t29389\t.\t+\t.\tName=NC_011744+29339..29389;\n+NC_011744\tsplit\tgene\t29390\t29614\t.\t+\t.\tName=NC_011744+29390..29614;\n+NC_011744\tsplit\tutr3\t29615\t29644\t.\t+\t.\tName=NC_011744+29615..29644;\n+NC_011744\tsplit\tutr5\t30066\t30126\t.\t+\t.\tName=NC_011744+30066..30126;\n+NC_011744\tsplit\tgene\t30127\t30615\t.\t+\t.\tName=NC_011744+30127..30615;\n+NC_011744\tsplit\tutr3\t30616\t30622\t.\t+\t.\tName=NC_011744+30616..30622;\n+NC_011744\tsplit\tutr5\t34131\t34144\t.\t+\t.\tName=NC_011744+34131..34144;\n+NC_011744\tsplit\tgene\t34145\t35119\t.\t+\t.\tName=NC_011744+34145..35119;\n+NC_011744\tsplit\tutr3\t35120\t35153\t.\t+\t.\tName=NC_011744+35120..35153;\n+NC_011744\tsplit\tutr5\t40845\t40886\t.\t+\t.\tName=NC_011744+40845..40886;\n+NC_011744\tsplit\tgene\t40887\t41426\t.\t+\t.\tName=NC_011744+40887..41426;\n+NC_011744\tsplit\tgene\t43855\t45801\t.\t+\t.\tName=NC_011744+43855..45801;\n+NC_011744\tsplit\tutr5\t45972\t46006\t.\t+\t.\tName=NC_011744+45972..46006;\n+NC_011744\tsplit\tgene\t46007\t49254\t.\t+\t.\tName=NC_011744+46007..49254;\n+NC_011744\tsplit\tgene\t49412\t51862\t.\t+\t.\tName=NC_011744+49412..51862;\n+NC_011744\tsplit\tgene\t57403\t58506\t.\t+\t.\tName=NC_011744+57403..58506;\n+NC_011744\tsplit\tutr5\t58837\t58865\t.\t+\t.\tName=NC_011744+58837..58865;\n+NC_011744\tsplit\tgene\t58866\t60065\t.\t+\t.\tName=NC_011744+58866..60065;\n+NC_011744\tsplit\tgene\t61052\t63159\t.\t+\t.\tName=NC_011744+61052..63159;\n+NC_011744\tsplit\tgene\t63377\t67642\t.\t+\t.\tName=NC_011744+63377..67642;\n+NC_011744\tsplit\tgene\t68078\t75324\t.\t+\t.\tName=NC_011744+68078..75324;\n+NC_011744\tsplit\tgene\t75688\t76170\t.\t+\t.\tName=NC_011744+75688..76170;\n+NC_011744\tsplit\tgene\t76628\t77284\t.\t+\t.\tName=NC_011744+76628..77284;\n+NC_011744\tsplit\tgene\t78060\t78320\t.\t+\t.\tName=NC_011744+78060..78320;\n+NC_011744\tsplit\tgene\t78790\t82528\t.\t+\t.\tName=NC_011744+78790..82528;\n+NC_011744\tsplit\tgene\t82754\t84273\t.\t+\t.\tName=NC_011744+82754..84273;\n+NC_011744\tsplit\tgene\t86398\t88921\t.\t+\t.\tName=NC_011744+86398..88921;\n+NC_011744\tsplit\tgene\t92897\t93433\t.\t+\t.\tName=NC_011744+92897..93433;\n+NC_011744\tsplit\tgene\t93893\t94486\t.\t+\t.\tName=NC_011744+93893..94486;\n+NC_011744\tsplit\tgene\t94865\t96993\t.\t+\t.\tName=NC_011744+94865..96993;\n+NC_011744\tsplit\tgene\t97330\t98777\t.\t+\t.\tName=NC_011744+97330..98777;\n+NC_011744\tsplit\tgene\t103591\t104505\t.\t+\t.\tName=NC_011744+103591..104505;\n+NC_011744\tsplit\tgene\t105690\t106655\t.\t+\t.\tName=NC_011744+105690..106655;\n+NC_011744\tsplit\tgene\t106826\t109364\t.\t+\t.\tName=NC_011744+106826..109364;\n+NC_011744\tsplit\tgene\t109568\t110131\t.\t+\t.\tName=NC_011744+109568..110131;\n+NC_011744\tsplit\tgene\t111700\t112581\t.\t+\t.\tName=NC_011744+111700..112581;\n+NC_011744\tsplit\tgene\t115500\t116063\t.\t+\t.\tName=NC_011744+115500..116063;\n+NC_011744\tsplit\tgene\t116223\t117755\t.\t+\t.\tName=NC_011744+116223..117755;\n+NC_011744\tsplit\tgene\t118880\t120124\t.\t+\t.\tName=NC_011744+118880..120124;\n+NC_011744\tsplit\tgene\t120342\t121256\t.\t+\t.\tName=NC_011744+120342..121256;\n+NC_011744\tsplit\tgene\t123070\t123306\t.\t+\t.\tName=NC_011744+123070..123306;\n+NC_011744\tsplit\tgene\t129984\t130850\t.\t+\t.\tName=NC_011744+129984..130850;\n+NC_011744\tsplit\tgene\t132757\t133554\t.\t+\t.\tName=NC_011744+132757..133554;\n+NC_011744\tsplit\tgen'..b'178073\t179716\t.\t-\t.\tName=NC_011744-178073..179716;\n+NC_011744\tsplit\tgene\t173629\t177705\t.\t-\t.\tName=NC_011744-173629..177705;\n+NC_011744\tsplit\tgene\t167706\t169679\t.\t-\t.\tName=NC_011744-167706..169679;\n+NC_011744\tsplit\tgene\t165711\t167525\t.\t-\t.\tName=NC_011744-165711..167525;\n+NC_011744\tsplit\tgene\t164814\t165557\t.\t-\t.\tName=NC_011744-164814..165557;\n+NC_011744\tsplit\tgene\t163858\t164139\t.\t-\t.\tName=NC_011744-163858..164139;\n+NC_011744\tsplit\tgene\t162181\t163424\t.\t-\t.\tName=NC_011744-162181..163424;\n+NC_011744\tsplit\tgene\t160700\t161692\t.\t-\t.\tName=NC_011744-160700..161692;\n+NC_011744\tsplit\tgene\t144807\t156233\t.\t-\t.\tName=NC_011744-144807..156233;\n+NC_011744\tsplit\tgene\t135655\t136593\t.\t-\t.\tName=NC_011744-135655..136593;\n+NC_011744\tsplit\tgene\t130935\t132437\t.\t-\t.\tName=NC_011744-130935..132437;\n+NC_011744\tsplit\tgene\t129398\t129829\t.\t-\t.\tName=NC_011744-129398..129829;\n+NC_011744\tsplit\tgene\t126512\t128949\t.\t-\t.\tName=NC_011744-126512..128949;\n+NC_011744\tsplit\tgene\t125434\t126219\t.\t-\t.\tName=NC_011744-125434..126219;\n+NC_011744\tsplit\tgene\t123588\t125222\t.\t-\t.\tName=NC_011744-123588..125222;\n+NC_011744\tsplit\tgene\t121363\t122403\t.\t-\t.\tName=NC_011744-121363..122403;\n+NC_011744\tsplit\tgene\t117879\t118601\t.\t-\t.\tName=NC_011744-117879..118601;\n+NC_011744\tsplit\tgene\t114873\t115439\t.\t-\t.\tName=NC_011744-114873..115439;\n+NC_011744\tsplit\tgene\t113921\t114712\t.\t-\t.\tName=NC_011744-113921..114712;\n+NC_011744\tsplit\tgene\t112703\t113697\t.\t-\t.\tName=NC_011744-112703..113697;\n+NC_011744\tsplit\tgene\t110231\t111424\t.\t-\t.\tName=NC_011744-110231..111424;\n+NC_011744\tsplit\tgene\t104593\t105597\t.\t-\t.\tName=NC_011744-104593..105597;\n+NC_011744\tsplit\tgene\t99962\t103402\t.\t-\t.\tName=NC_011744-99962..103402;\n+NC_011744\tsplit\tgene\t98945\t99784\t.\t-\t.\tName=NC_011744-98945..99784;\n+NC_011744\tsplit\tgene\t93618\t93836\t.\t-\t.\tName=NC_011744-93618..93836;\n+NC_011744\tsplit\tgene\t88922\t92825\t.\t-\t.\tName=NC_011744-88922..92825;\n+NC_011744\tsplit\tgene\t85279\t86277\t.\t-\t.\tName=NC_011744-85279..86277;\n+NC_011744\tsplit\tgene\t84373\t84948\t.\t-\t.\tName=NC_011744-84373..84948;\n+NC_011744\tsplit\tgene\t77461\t77685\t.\t-\t.\tName=NC_011744-77461..77685;\n+NC_011744\tsplit\tgene\t60250\t60756\t.\t-\t.\tName=NC_011744-60250..60756;\n+NC_011744\tsplit\tgene\t54927\t57177\t.\t-\t.\tName=NC_011744-54927..57177;\n+NC_011744\tsplit\tutr5\t54568\t54590\t.\t-\t.\tName=NC_011744-54568..54590;\n+NC_011744\tsplit\tgene\t52087\t54567\t.\t-\t.\tName=NC_011744-52087..54567;\n+NC_011744\tsplit\tgene\t41423\t43678\t.\t-\t.\tName=NC_011744-41423..43678;\n+NC_011744\tsplit\tutr3\t41380\t41422\t.\t-\t.\tName=NC_011744-41380..41422;\n+NC_011744\tsplit\tgene\t39954\t40544\t.\t-\t.\tName=NC_011744-39954..40544;\n+NC_011744\tsplit\tutr3\t39921\t39953\t.\t-\t.\tName=NC_011744-39921..39953;\n+NC_011744\tsplit\tgene\t38102\t39724\t.\t-\t.\tName=NC_011744-38102..39724;\n+NC_011744\tsplit\tgene\t37057\t37881\t.\t-\t.\tName=NC_011744-37057..37881;\n+NC_011744\tsplit\tgene\t35271\t36858\t.\t-\t.\tName=NC_011744-35271..36858;\n+NC_011744\tsplit\tutr5\t33579\t33638\t.\t-\t.\tName=NC_011744-33579..33638;\n+NC_011744\tsplit\tgene\t31728\t33578\t.\t-\t.\tName=NC_011744-31728..33578;\n+NC_011744\tsplit\tutr3\t31715\t31727\t.\t-\t.\tName=NC_011744-31715..31727;\n+NC_011744\tsplit\tgene\t30726\t31430\t.\t-\t.\tName=NC_011744-30726..31430;\n+NC_011744\tsplit\tutr5\t29848\t29874\t.\t-\t.\tName=NC_011744-29848..29874;\n+NC_011744\tsplit\tgene\t29677\t29847\t.\t-\t.\tName=NC_011744-29677..29847;\n+NC_011744\tsplit\tutr3\t29672\t29676\t.\t-\t.\tName=NC_011744-29672..29676;\n+NC_011744\tsplit\tutr5\t22525\t22542\t.\t-\t.\tName=NC_011744-22525..22542;\n+NC_011744\tsplit\tgene\t22321\t22524\t.\t-\t.\tName=NC_011744-22321..22524;\n+NC_011744\tsplit\tutr5\t21962\t22087\t.\t-\t.\tName=NC_011744-21962..22087;\n+NC_011744\tsplit\tgene\t20840\t21961\t.\t-\t.\tName=NC_011744-20840..21961;\n+NC_011744\tsplit\tutr3\t20746\t20839\t.\t-\t.\tName=NC_011744-20746..20839;\n+NC_011744\tsplit\tutr5\t20683\t20714\t.\t-\t.\tName=NC_011744-20683..20714;\n+NC_011744\tsplit\tgene\t18853\t20682\t.\t-\t.\tName=NC_011744-18853..20682;\n+NC_011744\tsplit\tgene\t17694\t18476\t.\t-\t.\tName=NC_011744-17694..18476;\n+NC_011744\tsplit\tutr5\t16788\t16790\t.\t-\t.\tName=NC_011744-16788..16790;\n+NC_011744\tsplit\tgene\t7180\t16787\t.\t-\t.\tName=NC_011744-7180..16787;\n'
b
diff -r 000000000000 -r a53eb951b164 detrprok_scripts/test-data/splitTranscript_TranscriptionUnit.gff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/detrprok_scripts/test-data/splitTranscript_TranscriptionUnit.gff Mon Mar 25 05:39:29 2013 -0400
b
b'@@ -0,0 +1,881 @@\n+NC_011744\tS-MART\ttranscript\t364\t4831\t.\t+\t.\tnbElements=334.000000;overlapsWith=YP_002394608.1--YP_002394607.1--YP_002394606.1;nbOverlaps=3;quality=255;ID=HWUSI-EAS1656_0001:3:89:5014:9610#0/1--HWUSI-EAS1656_0001:3:35:9162:10538#0/1--HWUSI-EAS1656_0001:3:;Name=HWUSI-EAS1656_0001:3:89:5014:9610#0/1--HWUSI-EAS1656_0001:3:35:9162:10538#0/1--HWUSI-EAS1656_0001:3:\n+NC_011744\tS-MART\ttranscript\t5319\t7032\t.\t+\t.\tnbElements=269.000000;overlapsWith=YP_002394609.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:107:12063:21287#0/1--HWUSI-EAS1656_0001:3:83:3280:18655#0/1--HWUSI-EAS1656_0001;Name=HWUSI-EAS1656_0001:3:107:12063:21287#0/1--HWUSI-EAS1656_0001:3:83:3280:18655#0/1--HWUSI-EAS1656_0001\n+NC_011744\tS-MART\ttranscript\t7180\t16790\t.\t-\t.\tnbElements=716.000000;overlapsWith=YP_002394615.1--YP_002394614.1--YP_002394613.1--YP_002394612.1--YP_002394611.1--YP_002394610.1;nbOverlaps=6;quality=255;ID=HWUSI-EAS1656_0001:3:80:8700:6553#0/1;Name=HWUSI-EAS1656_0001:3:80:8700:6553#0/1--HWUSI-EAS1656_0001:3:17:7827:9031#0/1--HWUSI-EAS1656_0001:3:2\n+NC_011744\tS-MART\ttranscript\t16915\t17644\t.\t+\t.\tnbElements=67.000000;overlapsWith=YP_002394616.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:72:18710:9619#0/1--HWUSI-EAS1656_0001:3:35:17678:14651#0/1;Name=HWUSI-EAS1656_0001:3:72:18710:9619#0/1--HWUSI-EAS1656_0001:3:35:17678:14651#0/1--HWUSI-EAS1656_0001:\n+NC_011744\tS-MART\ttranscript\t17694\t18476\t.\t-\t.\tnbElements=10.000000;overlapsWith=YP_002394617.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:27:15227:6678#0/1;Name=HWUSI-EAS1656_0001:3:27:15227:6678#0/1--HWUSI-EAS1656_0001:3:81:9376:6532#0/1--HWUSI-EAS1656_0001:3:\n+NC_011744\tS-MART\ttranscript\t18853\t20714\t.\t-\t.\tnbElements=822.000000;overlapsWith=YP_002394618.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:120:10699:20912#0/1--HWUSI-EAS1656_0001:3:98:17053:15919#0/1--HWUSI-EAS1656_000;Name=HWUSI-EAS1656_0001:3:120:10699:20912#0/1--HWUSI-EAS1656_0001:3:98:17053:15919#0/1--HWUSI-EAS1656_000\n+NC_011744\tS-MART\ttranscript\t20746\t22087\t.\t-\t.\tnbElements=1404.000000;overlapsWith=YP_002394619.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:105:17341:13775#0/1--HWUSI-EAS1656_0001:3:92:5836:8090#0/1--HWUSI-EAS1656_0001:;Name=HWUSI-EAS1656_0001:3:105:17341:13775#0/1--HWUSI-EAS1656_0001:3:92:5836:8090#0/1--HWUSI-EAS1656_0001:\n+NC_011744\tS-MART\ttranscript\t22321\t22542\t.\t-\t.\tnbElements=7.000000;overlapsWith=YP_002394620.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:51:12735:9404#0/1--HWUSI-EAS1656_0001:3:40:1950:13769#0/1--HWUSI-EAS1656_0001:3;Name=HWUSI-EAS1656_0001:3:51:12735:9404#0/1--HWUSI-EAS1656_0001:3:40:1950:13769#0/1--HWUSI-EAS1656_0001:3\n+NC_011744\tS-MART\ttranscript\t22423\t23272\t.\t+\t.\tnbElements=29.000000;overlapsWith=YP_002394621.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:117:18154:8751#0/1--HWUSI-EAS1656_0001:3:79:2790:12435#0/1--HWUSI-EAS1656_0001:;Name=HWUSI-EAS1656_0001:3:117:18154:8751#0/1--HWUSI-EAS1656_0001:3:79:2790:12435#0/1--HWUSI-EAS1656_0001:\n+NC_011744\tS-MART\ttranscript\t23395\t29212\t.\t+\t.\tnbElements=4357.000000;overlapsWith=YP_002394624.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:7:4762:13462#0/1--HWUSI-EAS1656_0001:3:75:10747:17479#0/1--HWUSI-EAS1656_0001:3;Name=HWUSI-EAS1656_0001:3:7:4762:13462#0/1--HWUSI-EAS1656_0001:3:75:10747:17479#0/1--HWUSI-EAS1656_0001:3\n+NC_011744\tS-MART\ttranscript\t29339\t29644\t.\t+\t.\tnbElements=32.000000;overlapsWith=YP_002394625.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:93:17472:16781#0/1--HWUSI-EAS1656_0001:3:85:8946:6720#0/1--HWUSI-EAS1656_0001:3;Name=HWUSI-EAS1656_0001:3:93:17472:16781#0/1--HWUSI-EAS1656_0001:3:85:8946:6720#0/1--HWUSI-EAS1656_0001:3\n+NC_011744\tS-MART\ttranscript\t29672\t29874\t.\t-\t.\tnbElements=167.000000;overlapsWith=YP_002394626.1;nbOverlaps=1;quality=255;ID=HWUSI-EAS1656_0001:3:95:2533:10057#0/1--HWUSI-EAS1656_0001:3:72:1679:20601#0/1--HWUSI-EAS1656_0001:3;Name=HWUSI-EAS1656_0001:3:95:2533:10057#0/1--HWUSI-EAS1656_0001:3:72:1679:20601#0/1--HWUSI-EAS1656_0001:3\n+NC_011744\tS-MART\ttranscript\t30066\t30622\t'..b'bkey=CDS;protein_id=YP_002396063.1;Dbxref=Genbank:YP_002396063.1,GeneID:7138770;ID=cds1457;Name=YP_002396063.1--YP_002396062.1\n+NC_011744\tS-MART\tCDS\t1645252\t1646514\t.\t+\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396064.1;Dbxref=Genbank:YP_002396064.1,GeneID:7138771;ID=cds1458;Name=YP_002396064.1\n+NC_011744\tS-MART\tCDS\t1646662\t1649029\t.\t-\t.\tproduct=phenylalanine 4-monooxygenase;nbElements=3.000000;transl_table=11;Note=phenylalanine 4-hydroxylase%3B phenylalanine 4-hydroxylase%3B catalyzes the formation of 4a-hydroxytetrahydrobiopterin and tyrosine from phenylalanine and tetrahydrobiopterin;gbkey=CDS;protein_id=YP_002396067.1;Dbxref=Genbank:YP_002396067.1,GeneID:7138774;ID=cds1461;Name=YP_002396067.1--YP_002396066.1--YP_002396065.1\n+NC_011744\tS-MART\tCDS\t1649229\t1651280\t.\t+\t.\tproduct=acetoacetyl-CoA synthetase;transl_table=11;Note=AcsA%3B in Sinorhizobium meliloti this enzyme is required for acetoacetate activation%3B similar to acetyl-CoA synthase;gbkey=CDS;protein_id=YP_002396068.1;Dbxref=Genbank:YP_002396068.1,GeneID:7138775;ID=cds1462;Name=YP_002396068.1\n+NC_011744\tS-MART\tCDS\t1651672\t1657718\t.\t+\t.\tproduct=hypothetical protein;nbElements=6.000000;transl_table=11;gbkey=CDS;protein_id=YP_002396074.1;Dbxref=Genbank:YP_002396074.1,GeneID:7138781;ID=cds1468;Name=YP_002396074.1--YP_002396073.1--YP_002396072.1--YP_002396071.1--YP_002396070.1--YP_002396069.1\n+NC_011744\tS-MART\tCDS\t1658632\t1659561\t.\t+\t.\tproduct=carbamate kinase;transl_table=11;Note=reversible synthesis of carbamate and ATP from carbamoyl phosphate and ADP;gbkey=CDS;protein_id=YP_002396075.1;Dbxref=Genbank:YP_002396075.1,GeneID:7138782;ID=cds1469;Name=YP_002396075.1\n+NC_011744\tS-MART\tCDS\t1659797\t1663769\t.\t-\t.\tproduct=4-hydroxyphenylpyruvate dioxygenase;nbElements=4.000000;transl_table=11;Note=Evidence 2a : Function of homologous gene experimentally demonstrated in another organism%3B PubMedId : 8112844%3B Product type e : enzyme;gbkey=CDS;protein_id=YP_002396079.1;Dbxref=Genbank:YP_002396079.1,GeneID:7138786;ID=cds1473;Name=YP_002396079.1--YP_002396078.1--YP_002396077.1--YP_002396076.1\n+NC_011744\tS-MART\tCDS\t1663853\t1664035\t.\t+\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396080.1;Dbxref=Genbank:YP_002396080.1,GeneID:7138787;ID=cds1474;Name=YP_002396080.1\n+NC_011744\tS-MART\tCDS\t1664130\t1664789\t.\t-\t.\tproduct=phosphoserine phosphatase;transl_table=11;gbkey=CDS;protein_id=YP_002396081.1;Dbxref=Genbank:YP_002396081.1,GeneID:7138788;ID=cds1475;Name=YP_002396081.1\n+NC_011744\tS-MART\tCDS\t1665142\t1667568\t.\t-\t.\tproduct=two-component sensor;transl_table=11;gbkey=CDS;protein_id=YP_002396082.1;Dbxref=Genbank:YP_002396082.1,GeneID:7138789;ID=cds1476;Name=YP_002396082.1\n+NC_011744\tS-MART\tCDS\t1667657\t1668415\t.\t+\t.\tproduct=amino acid ABC transporter substrate-binding protein;transl_table=11;gbkey=CDS;protein_id=YP_002396083.1;Dbxref=Genbank:YP_002396083.1,GeneID:7138790;ID=cds1477;Name=YP_002396083.1\n+NC_011744\tS-MART\tCDS\t1668625\t1669185\t.\t-\t.\tproduct=GTP cyclohydrolase II;transl_table=11;gbkey=CDS;protein_id=YP_002396084.1;Dbxref=Genbank:YP_002396084.1,GeneID:7138791;ID=cds1478;Name=YP_002396084.1\n+NC_011744\tS-MART\tCDS\t1669376\t1669699\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396085.1;Dbxref=Genbank:YP_002396085.1,GeneID:7138792;ID=cds1479;Name=YP_002396085.1\n+NC_011744\tS-MART\tCDS\t1669923\t1672240\t.\t+\t.\tproduct=hypothetical protein;nbElements=3.000000;transl_table=11;gbkey=CDS;protein_id=YP_002396088.1;Dbxref=Genbank:YP_002396088.1,GeneID:7138795;ID=cds1482;Name=YP_002396088.1--YP_002396087.1--YP_002396086.1\n+NC_011744\tS-MART\tCDS\t1672182\t1672271\t.\t-\t.\tproduct=hypothetical protein;transl_table=11;gbkey=CDS;protein_id=YP_002396089.1;Dbxref=Genbank:YP_002396089.1,GeneID:7138796;ID=cds1483;Name=YP_002396089.1\n+NC_011744\tS-MART\tCDS\t1672683\t1674662\t.\t-\t.\tproduct=RctB protein;transl_table=11;gbkey=CDS;protein_id=YP_002396090.1;Dbxref=Genbank:YP_002396090.1,GeneID:7138797;ID=cds1484;Name=YP_002396090.1\n'