Changeset 269

Timestamp:

09/24/09 11:42:35 (2 months ago)

Author:

cholt

Message:

try to fix weird randon cwd regular expression related error

Files:

• lib/GI.pm (modified) (2 diffs)
• lib/ds_utility.pm (modified) (1 diff)
• lib/runlog.pm (modified) (5 diffs)

lib/GI.pm

@@ -606 +606 @@
-      #fix non standard peptides
-      if (defined $alt) {
-\*
+[\*\-]
-         $$seq_ref =~ s/[^abcdefghiklmnpqrstvwyzxABCDEFGHIKLMNPQRSTVWYZX\-\n]/$alt/g;
+      }
+      #fix nucleotide sequences
+      elsif($key !~ /protein/){
+          #most programs use N for masking but for some reason the NCBI decided to
+          #use X to mask their sequence, which causes many many programs to fail
+          $$seq_ref =~ s/\-//g;
+          $$seq_ref =~ s/X/N/g;
+          die "ERROR: The nucleotide sequence file \'$file\'\n".
+              "appears to contain protein sequence or unrecognized characters.\n".
+              "Please check/fix the file before continuing.\n".
+              "Invalid Character: $1\n\n"
+              if($$seq_ref =~ /([^acgturykmswbdhvnxACGTURYKMSWBDHVNX\-\n])/);
-      }
-
@@ -2237 +2249 @@
-      $CTL_OPT{'single_length'} = 250;
-      $CTL_OPT{'min_protein'} = 0;
-
+l
-      $CTL_OPT{'keep_preds'} = 0;
-      $CTL_OPT{'map_forward'} = 0;

lib/ds_utility.pm

@@ -127 +127 @@
-
-   #remove deep directory data so log is relative
-
+get
-   my $entry = join("\t", @F);
+
+   #while loop is used to solve weird incorrect cwd
+   #that happens randomly on the cluster
+   while($cwd ne Cwd::getcwd()){
+       $cwd = Cwd::getcwd();
+   }
-
-   if($entry =~ /\tFINISHED|\tSTARTED|\tDIED|\tSKIPPED|\tRETRY/){

lib/runlog.pm

@@ -197 +197 @@
-        if($continue_flag >= 0 || $continue_flag == -1){
-            #CHECK CONTROL FILE OPTIONS FOR CHANGES
-
-
+
+
+
+
+
+
+
+
+
-                #these are only sometimes important
-                if($key =~ /^est_pass$|^altest_pass$|^protein_pass$|^rm_pass$/ ||
@@ -204 +211 @@
-                   ){
-                    next unless($CTL_OPTIONS{genome_gff});
-
+
+
-                    $old =~ s/^$cwd\/*//;
-                    my $new = $CTL_OPTIONS{genome_gff};
-                    $new =~ s/^$cwd\/*//;
+
+                    #temp lamprey
+                    my $chpc = "scratch/serial-pio/u0045039/lamprey2/";
+                    $old =~ s/^$chpc\/*//;
+                    $new =~ s/^$chpc\/*//;
+                    $chpc = "scratch/serial/u0045039/lamprey2/";
+                    $old =~ s/^$chpc\/*//;
+                    $new =~ s/^$chpc\/*//;
+
-
-                    #only continue if change not already happening
@@ -365 +382 @@
-                        "did not finish on the last run and must be erased\n";
-                    push(@files, $key);
+
+                    #this next step will get both temp directories and files that
+                    #have the incorrect location in the log but are in theVoid
-                    $key =~ /([^\/]+)$/;
-                    my $rm_f_name = $1;
+
+                    if(! -e $key && -e "$the_void/$rm_f_name"){
+                        push(@files, $rm_f_name);
+                    }
+
-                    $rm_f_name =~ s/\.fasta$//;
-                    my @d = <$the_void/*$rm_f_name*>;
@@ -589 +614 @@
-
-   #log control file options
-
+
+
+
+
+
+
+
+
-   foreach my $key (@ctl_to_log) {
-      my $ctl_val = '';
@@ -618 +650 @@
-
-   my $log_file = $self->{file_name};
-
+
+
+
+
+
+
+
-
-   #this line hides unnecessarilly deep directory details
-   #this is important for maker webserver security
+   #also important when moving work directory around
-   if(defined $key && defined $type){
-       $key =~ s/^$cwd\/*// if($type =~ /^STARTED$|^FINISHED$/);