JPRED-2 Initial commit of software for the Jpred website (some files excluded due...

[jpred.git] / websoft / bin / concise2blc

diff --git a/websoft/bin/concise2blc b/websoft/bin/concise2blc
new file mode 100755 (executable)
index 0000000..f22c66a
--- /dev/null
+++ b/websoft/bin/concise2blc
@@ -0,0 +1,114 @@
+#!/usr/bin/perl
+
+#
+# Generates a blc file from the concise file, reads in the sequence
+# alignments those records in @data
+#
+
+#
+#  CC 22/06/07 - bug fix in 'conversion' of coil-coiled predictions
+#  CC 22/05/06 - bug fix in final foreach loop
+#
+
+use strict;
+use warnings;
+
+if ($ARGV[0]) { open(IN, "<$ARGV[0]") or die($!); }
+else { open(IN, "<-") or die($!); }
+
+# The names of the records from the concise file that we want
+my @data = ("JNETALIGN", "JNETHMM", "jnetpred", "JNETPSSM", "JNETCONF", "JNETSOL25", "JNETSOL5", "JNETSOL0", "Lupas_21", "Lupas_14", "Lupas_28");
+
+my %convert = ("JNETALIGN" => "jalign",
+       "JNETHMM" => "jhmm",
+       "jnetpred" => "jnet",
+       "JNETPSSM" => "jpssm",
+       "JNETCONF" => "conf",
+       "JNETSOL25" => "sol25",
+       "JNETSOL5" => "sol5",
+       "JNETSOL0" => "sol0",
+       "Lupas_21" => "lupas_21",
+       "Lupas_14" => "lupas_14",
+       "Lupas_28" => "lupas_28",
+       );
+
+my (@seq, %seq, @pred, %pred);   # CC - these could be replaced with a tied hash, for future edits.
+
+#
+# Read in the concise file and extract the data
+#
+
+while (<IN>) {
+       if (/^\n/) { next; }
+       my ($id, $seq) = split(":", $_);
+       if (!$id || !$seq) { next; }            # Check we have proper values
+       chomp($seq);
+       $seq =~ s/,//g;
+       if ($id =~ /align\d*;/) {
+               @_ = split(";", $id);           # Then its an alignment
+               push @seq, $_[1];
+               $seq{$_[1]} = $seq;
+               }
+       foreach (@data) {
+               if ($id eq $_) {
+                       push @pred, $_;
+                       $pred{$_} = $seq;
+                       }
+               }
+       }
+close(IN);
+
+my @blc;
+my $gap;
+my $seq_len = length($seq{$seq[0]}) - 1; 
+foreach (0..$seq_len) { $gap .= " "; }
+
+push @blc, $gap;
+
+foreach (@seq) {
+       $seq{$_} =~ s/\.|-/ /g;
+       push @blc, $seq{$_};
+       }
+push @blc, $gap;
+push @blc, $gap;
+push @blc, $gap;
+foreach (@pred) {
+       ## CC 22/06/07 - Convert 8-state secondary structure types to 3-state types unless
+       ## it's a coil-coil prediction where 'C' is a valid prediction.
+       ## I think this is a throw-back from when Jpred was a consensus server?
+   if ($_ =~ /Lupas/i) {
+      $pred{$_} =~ s/-/ /g;
+   } else {
+      $pred{$_} =~ s/[TCYWXZ_SI\?-]/ /g;
+      $pred{$_} =~ s/G/H/g;
+   }
+   push @blc, $pred{$_};
+}
+
+foreach (@seq) {
+       print ">$_\n";
+       }
+print ">\n>\n>\n";
+foreach (@pred) {
+       print ">$convert{$_}\n";
+       }
+print " * iteration 1\n";
+#print " $seq_len\n";
+
+foreach my $i (0..$seq_len) {
+       foreach (@blc) {
+      ### CC 22/05/06 - below has been changed due to bug where if
+      # statement return false when $b == 0. This should actually
+      # be true when parsing the JNETCONF string of integers.
+      #
+      # Now we test whether the variable is defined rather than true.
+      # NB: not sure what do if the variable is not defined?
+      
+               #if (my $b = substr($_, $i, 1)) { print $b; }
+      
+      my $value = substr($_, $i, 1);
+      print $value if defined($value);
+               }
+       print "\n";
+       }
+print " *";