Update to CM profiles and fixed a small bug with get_unique_features.pl

NGS-Utils/README.pod

@@ -831,6 +831,6 @@ This program is distributed under the Artistic License 2.0.
 
 =head1 DATE
 
-Feb-07-2018
+Feb-12-2018
 
 =cut

NGS-Utils/annotate_final_ncRNAs.sh

@@ -2,12 +2,12 @@
 
 # (C) Kranti Konganti
 # This program is distributed as Artistic License 2.0.
-# 11/15/2017
+# 02/12/2018
 # Coordinate with lncRNApipe output to parse out and add Infernal annotation to final ncRNA transcripts.
 
 # $LastChangedBy: konganti $ =~ m/.+?\:(.+)/;
-# $LastChangedDate: 2017-11-15 14:45:27 -0500 (Wed, 15 November 2017)  $ =~ m/.+?\:(.+)/;
-# $LastChangedRevision: 2707 $ =~ m/.+?\:\s*(.*)\s*.*/;
+# $LastChangedDate: 2018-02-12 09:45:27 -0500 (Mon, 12 February 2018)  $ =~ m/.+?\:(.+)/;
+# $LastChangedRevision: 2708 $ =~ m/.+?\:\s*(.*)\s*.*/;
 # $AUTHORFULLNAME = 'Kranti Konganti';
 
 if [  -z "$FINAL_GTF" ]  ||
@@ -26,7 +26,8 @@ grep noncoding $CPC_TXT_OUT | cut -f 1 | sort -n | uniq | while read unetrid; do
     contig_en=`grep -P "\ttranscript\t.+?\"$trid\".+" $FINAL_GTF | awk '{print $5}'`;
     trlen=`grep -P "\"$trid\"" $FINAL_GTF | grep -oP 'transcript_length \"\d+\"' | head -n 1 | perl -e '\$line = <>; if (\$line =~ m/.+?(\d+)/) {print \$1;}'`;
     hitlen=`grep -P "\s+$trid\s+" $CM_TXT_OUT | grep -P "\\s+\!\\s+" | sort -k15,15nr | uniq | head -n 1 | awk '{if(\$10=="+") print \$9-\$8; else print \$8-\$9;}'`;
-    annot=`grep -P "\s+$trid\s+" $CM_TXT_OUT | grep -P "\\s+\!\\s+" | sort -k15,15nr | uniq | awk '{$1=""; sc=$15; for(i=3;i<=17;i++) $i=""; if (length($0) != 0) print $0, " | BitScore: ",sc}' | cut -d " " -f 2- | sed -e 's/\s\+/ /g' | head -n 1`;
+    annot=`grep -P "\s+$trid\s+" $CM_TXT_OUT | grep -P "\\s+\!\\s+" | sort -k15,15nr | uniq | awk '{sc=$15; for(i=1;i<=17;i++); if (length($0) != 0) print "Match: ", $1, " | BitScore: ", sc}' | sed -e 's/\s\+/ /g' | head -n 1`;
+    #annot=`grep -P "\s+$trid\s+" $CM_TXT_OUT | grep -P "\\s+\!\\s+" | sort -k15,15nr | uniq | awk '{$1=""; sc=$15; for(i=3;i<=17;i++) $i=""; if (length($0) != 0) print $0, " | BitScore: ",sc}' | cut -d " " -f 2- | sed -e 's/\s\+/ /g' | head -n 1`;
 
     if [ -z "$trlen" ] || [ -z "$hitlen" ]; then
 	calc_cov=0.0;

NGS-Utils/get_unique_features.pl

@@ -10,8 +10,8 @@
 use Set::IntervalTree;
 
 my ($LASTCHANGEDBY) = q$LastChangedBy: konganti $ =~ m/.+?\:(.+)/;
-my ($LASTCHANGEDDATE) = q$LastChangedDate: 2016-01-11 12:48:27 -0500 (Mon, 1 Jan 2016)  $ =~ m/.+?\:(.+)/;
-my ($VERSION) = q$LastChangedRevision: 0605 $ =~ m/.+?(\d+)/;
+my ($LASTCHANGEDDATE) = q$LastChangedDate: 2018-02-12 10:48:27 -0500 (Mon, 12 Feb 2018)  $ =~ m/.+?\:(.+)/;
+my ($VERSION) = q$LastChangedRevision: 0606 $ =~ m/.+?(\d+)/;
 my $AUTHORFULLNAME = 'Kranti Konganti';
 
 my $io = IO::Routine->new();
@@ -199,6 +199,8 @@ sub compare_feat {
     while (my $line = <$s_fh>) {
 
 	chomp $line;
+	next if ($line =~ m/^\W/);
+
 	my ($left_coords, $right_coords) = [];
 
 	$line = $io->strip_leading_and_trailing_spaces($line);
@@ -261,6 +263,8 @@ sub compare_feat {
     my $insert_line = 0;
     while (my $line = <$c_fh>) {
 	chomp $line;
+	next if ($line =~ m/^\W/);
+
 	$line = $io->strip_leading_and_trailing_spaces($line);
 
 	my @cols = split/\t/, $line;
@@ -653,6 +657,6 @@ =head1 COPYRIGHT
 
 =head1 DATE
 
-Jan-11-2016
+Feb-12-2018
 
 =cut

NGS-Utils/lncRNApipe

@@ -12,8 +12,8 @@ use Data::Dumper;
 Getopt::Long::Configure('prefix=--');
 
 my ($LASTCHANGEDBY) = q$LastChangedBy: konganti $ =~ m/.+?\:(.+)/;
-my ($LASTCHANGEDDATE) = q$LastChangedDate: 2018-02-07 09:27:27 -0500 (Wed, 07 February 2018)  $ =~ m/.+?\:(.+)/;
-my ($VERSION) = q$LastChangedRevision: 1.2.1 $ =~ m/.+?\:\s*(.*)\s*.*/;
+my ($LASTCHANGEDDATE) = q$LastChangedDate: 2018-02-12 09:27:27 -0500 (Mon, 12 February 2018)  $ =~ m/.+?\:(.+)/;
+my ($VERSION) = q$LastChangedRevision: 1.2.2 $ =~ m/.+?\:\s*(.*)\s*.*/;
 my $AUTHORFULLNAME = 'Kranti Konganti';
 
 my ($help, $quiet, $setup, $get_uq_sc_opts,
@@ -822,6 +822,11 @@ sub run_lncRNApipe {
 
     # \x{2633}
     if ($lncRNApipe_succ) {
+	$io->warning('INFERNAL authors state the following:' . 
+		     "\n\n" . 'We recommend that you use Rfam/Infernal for vertebrate genome annotation with extreme caution! ' .
+		     "\n" . 'Nevertheless, Rfam/Infernal does tell us about important sequence similarities that are effectively undetectable by other means. ' .
+		     "\n" . 'However, in complex eukaryotic genomes, it is important to treat hits as sequence similarity ' .
+		     "\n" . 'information (much as you might treat BLAST hits), rather than as evidence of bona fide ncRNA genes.', 'INFO!!');
 	$io->c_time("\x{2632}\x{2634} lncRNApipe Pipeline finished!");
     }
     else {
@@ -3862,6 +3867,6 @@ This program is distributed under the Artistic License 2.0.
 
 =head1 DATE
 
-Feb-07-2018
+Feb-12-2018
 
 =cut