ProtoGene.pl 62.9 KB
Newer Older
Sebastien Moretti's avatar
Sebastien Moretti committed
1
2
3
4
5
6
7
8
9
10
11
12
#!/usr/bin/env perl

########################################################
#                                                      #
# ProtoGene : Bona-Fide Back Translation of Protein    #
#             Multiple Sequence Alignments             #
#                                                      #
########################################################

use strict;
use warnings;
use diagnostics;
Sebastien Moretti's avatar
Sebastien Moretti committed
13
14
use Carp;

Sebastien Moretti's avatar
Sebastien Moretti committed
15
use File::Which qw(which);                 # Locate external executable programs in the PATH
16
17
use Time::localtime;                       # Use localtime+PID for a pseudo-uniq temp file name
use Getopt::Long;                          # Options specifications
Sebastien Moretti's avatar
Sebastien Moretti committed
18
use File::Copy qw(move);                   # Avoid external 'mv' command usage
19
use LWP::Simple;                           # To test gigablaster availability
Sebastien Moretti's avatar
Sebastien Moretti committed
20
21
22

use Mail::Send;                            # Send warnings and errors files by e-mail ==> only if the $userEMail variable is defined

23
24
use lib '/mnt/common/share/ProtoGene/';    # Local path for ProtoGene's own perl modules
###use lib '/mnt/local/lib/tcoffee_perl/';
Sebastien Moretti's avatar
Sebastien Moretti committed
25
use Exonerate;                             # Exonerate runner, parser, ...
26
use Views;                                 # Non-text outputs, e.g. HTML/CSS
27
#use CheckOutput;                         # Check output for cds consistancy with query
Sebastien Moretti's avatar
Sebastien Moretti committed
28
29
30



31
################## CONFIGURATION ##################
Sebastien Moretti's avatar
Sebastien Moretti committed
32
$ENV{'PATH'}        .= ':/mnt/local/bin/:./';                               # Additional path for executables
33
34

my $cachePath        = '/scratch/cluster/monthly/t_coffee/ProtoGene_Cache'; # Cache directory
35
$cachePath = '/Users/smoretti/Documents/ProtoGene/TTMMPP'; #FIXME
36
37
38
my $cacheStorageTime = 15;                                                  # Do not update sequences younger than X days

my $userEMail        = 'moretti.sebastien@gmail.com';                       # To receive e-mails with encountered problems; leave blank to inactive
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55

### BLAST parameters ###
my $blast_param = { 'evalue'    => 0.05,
                    'matrix'    => 'BLOSUM62',
                    'filter'    => 'Off',
                    'species'   => 'All_organisms',

                    'db1'       => 'refseq_protein', # NCBI RefSeq protein
                    'identity1' => 100,
                    'cover1'    => 100,
                    'db2'       => 'nr', # NCBI nr protein
                    'identity2' =>  95,
                    'cover2'    =>  95,

                    'nbesthits' => 10,
                    'core'      => 1,
                  };
56
###################################################
Sebastien Moretti's avatar
Sebastien Moretti committed
57
58


59
my $VERSION  = '4.0.0';
Sebastien Moretti's avatar
Sebastien Moretti committed
60

Sebastien Moretti's avatar
Sebastien Moretti committed
61
my $webblast_exe  = 'webblast.pl';
62
my $blast_exe     = 'blastall';            # Or wu-blastall for Wu-BLAST; for local blast usage
Sebastien Moretti's avatar
Sebastien Moretti committed
63
my $exonerate_exe = 'exonerate-1.0';       # Exonerate 1.0 because current parser only works with this version
64
65
66
67
68


################## Option management
my ($msa, $revtrans, $pep, $hideBOJ, $run_name, $template, $lim, $cache) = ('', 0, 0, 0, '', '', 0, 'update');
my ($debug, $tmp)                                                        = (0, 0);
69
my ($db, $species, $local, $giga)                                        = ($blast_param->{'db1'}, $blast_param->{'species'}, 0, 0);
Sebastien Moretti's avatar
Sebastien Moretti committed
70
71
72
73
74
75
76
77
78
79
80
my %opts = ('msa|in=s'        => \$msa,        # Input sequences
            'revtrans'        => \$revtrans,   # Use to reverse-translate sequences with no match
            'pep'             => \$pep,        # Add the original peptide query beneath the related CDS seq
            'hideBOJ'         => \$hideBOJ,    # Hide BOJ output
            'run_name=s'      => \$run_name,   # Use another name, instead of input seq name, for result files
            'template=s'      => \$template,   # Use a template file
            'lim=i'           => \$lim,        # Limit number of input query sequences
            'cache=s'         => \$cache,      # Cache behavior

            'orgm|species=s'  => \$species,    # Organism(s) to blast against
            'db|database=s'   => \$db,         # Database to blast against
81
            'local'           => \$local,      # Use to specify a local db query, defined in $local_db
Sebastien Moretti's avatar
Sebastien Moretti committed
82
83
84
85
86
87
88
89
90
91
            'giga'            => \$giga,       # Use GigaBlaster server

            'version|V'       => sub { print "\n\tPROTOGENE : Bona-Fide Back Translation of Protein Multiple Sequence Alignments\n\tversion   : $VERSION\n\n";
                                       exit 0;
                                 },            # Print version information
            'help|H|?'        => sub { system('perldoc', "$0")==0 || print {*STDERR} "\n\tCannot open the documentation with 'perldoc'\n\n";
                                       exit 0;
                                     },        # Print full help message
            'debug'           => \$debug,      # Verbose output
            'tmp'             => \$tmp,        # To keep traces of fake intermediate files like fake xml from NCBI, fake aln, ...
Sebastien Moretti's avatar
Sebastien Moretti committed
92
           );
Sebastien Moretti's avatar
Sebastien Moretti committed
93
94
my $test_option_values = Getopt::Long::GetOptions(%opts);

95
96

################## Short help message
97
if ( !$test_option_values || ($msa eq '' && $cache ne 'empty' && $cache ne 'old') ){
98
99
100
101
102
103
104
105
    print {*STDERR} "\n\tCannot open the MSA file in FASTA format
\tTry:  $0 --msa=path_of_the_fasta_msa_file [Options]

\tOptions: --orgm=All_organisms, Bacteria, Viruses, Vertebrata,
\t                Eukaryota, Mammalia, Primates, Homo_sapiens,
\t                Gallus_gallus, Bos_taurus, Escherichia_coli,
\t                Arabidopsis_thaliana, Mus_musculus,
\t                Drosophila_Melanogaster, ...
106
\t                default is '$blast_param->{'species'}'
107
\t         --db=nr, pdb, swissprot, refseq_protein
108
\t                default is '$blast_param->{'db1'}'
109
110
111
112
113
114
115
116
117
118
119
120
121
122
\t         --local to execute a local BLAST query with
\t                --db=path_for_a_local_db_blast_formated\n
\t         --template to provide your own nucleotidic sequences
\t                following the cds file format
\t         --revtrans to reverse-translate sequences with no
\t                blast hit, in IUB (IUPAC) depiction code
\t                They are removed from the alignement by default
\t         --pep  to add the original peptide query beneath the
\t                back-translated sequence
\t         --cache=none, update, use, 'own_PATH_directory', old, empty
\t                to select the cache mode
\t                default is 'update'\n
\t         --version to print version information
\t         --help    to print a full help message\n\n";
123
    exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
124
}
125
126


127
128
################## Check cache directory accessibility
checkCacheAccessibility($cachePath);
129

130
131
# Cache management
$cache = cacheManagement($cache, $cachePath, 0) || '.';
132

133
134
# Remove old files from the cache directory everytime ProtoGene is running
cacheManagement('old', $cachePath, 1);
135
136


137
138

################## Check external software in the PATH
139
checkExternalSoftware($webblast_exe, $exonerate_exe);
140
141
if ( $local ){
    checkExternalSoftware($blast_exe);
Sebastien Moretti's avatar
Sebastien Moretti committed
142
143
144
145
}



146
147
148
################## Check input files
# Open and check template file
my ($template_NT, $template_AA, $template_SEQ);
149
#TODO: template is fasta only now
150
151
($template_NT, $template_AA, $template_SEQ) = checkTemplate($template) if ( $template );

Sebastien Moretti's avatar
Sebastien Moretti committed
152

Sebastien Moretti's avatar
Sebastien Moretti committed
153
154
155
156
157
# Open and Check the input fasta file
my $originalMSA    = $msa;
my $converted      = 0; # Accept only fasta format
#TODO: Do a better file converter and identifier is msa is not fasta
#($msa, $converted) = isFastaFile($msa, $converted);
Sebastien Moretti's avatar
Sebastien Moretti committed
158

159
my $fasta_checker = -1;
160
my (@original_names, @original_seq); #Use 2 lists in parallel : original_names, seq
Sebastien Moretti's avatar
Sebastien Moretti committed
161
162
open( my $MSA, '<', "$msa" ) or die "\n\tCannot open your MSA file\n\n";
READ_MSA:
163
while(<$MSA>){
Sebastien Moretti's avatar
Sebastien Moretti committed
164
165
    next READ_MSA if ( $_ =~ /^#/ );

166
    if ( $_ =~ /^>/ ){
167
168
        $fasta_checker++;
        my $name = $_;
Sebastien Moretti's avatar
Sebastien Moretti committed
169
170
171
        $name    =~ s{\r\n}{}g; #Remove return lines from windows OS '^M'
        $name    =~ s{\r}{\n}g;
        $name    =~ s{^>[ \t]+}{>};
172
        chomp($name);
Sebastien Moretti's avatar
Sebastien Moretti committed
173
174

        push @original_names, $name;
Sebastien Moretti's avatar
Sebastien Moretti committed
175
    }
Sebastien Moretti's avatar
Sebastien Moretti committed
176
177
178
179
180
181
182
183
184
185
    elsif ( $_ =~ /[A-Z\-\.]/i ){
        my $sequence = $_;
        $sequence    =~ s{\r\n}{}g;
        $sequence    =~ s{\r}{\n}g;
        $sequence    =~ s{\.}{-}g; #for msa with '.' as gap
        #TODO because exonerate cannot work with selenocystein
        $sequence    =~ s{[BJOUZ]}{X}ig; #U here for selenocystein
        $sequence    =~ s{[^A-Za-z\-\*\n]}{}g; #Remove all the non-gap or non-alphabetic characters from the seq
        chomp($sequence);

Sebastien Moretti's avatar
Sebastien Moretti committed
186
        #fasta sequence on 1 line
187
        if ( !exists($original_seq[$fasta_checker]) ){
Sebastien Moretti's avatar
Sebastien Moretti committed
188
            push @original_seq, $sequence;
189
190
        }
        else {
Sebastien Moretti's avatar
Sebastien Moretti committed
191
            $original_seq[$fasta_checker] .= $sequence;
192
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
193
194
    }
}
195
close $MSA;
Sebastien Moretti's avatar
Sebastien Moretti committed
196
197

#Are there FASTA sequences ?
198
if ( $fasta_checker == -1 ){
Sebastien Moretti's avatar
Sebastien Moretti committed
199
    failure();
Sebastien Moretti's avatar
Sebastien Moretti committed
200
201
    print {*STDERR} "\tThe MSA file does NOT seem to be a protein FASTA format
\tSee an example of FASTA format <a href='http://en.wikipedia.org/wiki/FASTA_format' target='_blank'>here</a>\n\n";
202
    exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
203
}
204
elsif ( $lim >0 && $fasta_checker > $lim ){
Sebastien Moretti's avatar
Sebastien Moretti committed
205
    failure();
206
    print {*STDERR} "\tThe FASTA file is too large, try with less than $lim sequences\n\tor split your file\n\n";
207
208
    exit(1);
}
209
elsif ( exists( $original_seq[0] ) && $original_seq[0] =~ /[acgtu]/i ){
210
    my $first_seq = $original_seq[0];
Sebastien Moretti's avatar
Sebastien Moretti committed
211
212
213
214
215
216
    #Check if sequences are amino acids and not nucleotides ONLY on the 1st sequence
    my $a         = ($first_seq =~ s{[aA]}{}g)               || 0;
    my $c         = ($first_seq =~ s{[cC]}{}g)               || 0;
    my $g         = ($first_seq =~ s{[gG]}{}g)               || 0;
    my $t         = ($first_seq =~ s{[tTuU]}{}g)             || 0;
    my $non       = ($first_seq =~ s{[^aAcCgGtTuUXxNn-]}{}g) || 0;
217
    if ( ($a+$c+$g+$t) >= (($a+$c+$g+$t+$non)*80/100) ){
Sebastien Moretti's avatar
Sebastien Moretti committed
218
        failure();
Sebastien Moretti's avatar
Sebastien Moretti committed
219
220
        print {*STDERR} "\tYour sequences seem already to be nucleotides
\tthe goal of this program is to turn AMINO ACID alignments into CDS nucleotide alignments\n\n";
221
222
        exit(1);
    }
Sebastien Moretti's avatar
Sebastien Moretti committed
223
224
225
226
}
undef $fasta_checker;


Sebastien Moretti's avatar
Sebastien Moretti committed
227
# Check GigaBlaster status if used
228
if ( $giga==1 ){
229
230
    $local = 0;
    $local = 2 if ( head('http://www.igs.cnrs-mrs.fr/Giga2/~database/remoteblast.cgi') );
231
232
}

Sebastien Moretti's avatar
Sebastien Moretti committed
233

Sebastien Moretti's avatar
Sebastien Moretti committed
234
# Temporary file extension
235
my $date = sprintf("%d-%02d-%02d_%02dh%02d", localtime->year() + 1900, localtime->mon(), localtime->mday(), localtime->hour, localtime->min);
Sebastien Moretti's avatar
Sebastien Moretti committed
236
237


238

Sebastien Moretti's avatar
Sebastien Moretti committed
239
#Start main program with version # of programs and list original queries
240
print "\n\t   Protogene\t$VERSION\t$date\n\n";
241

242
open(my $EXONERATEISHERE, "$exonerate_exe --version |");
243
my $IsExonerateHere = <$EXONERATEISHERE>;
244
close $EXONERATEISHERE;
Sebastien Moretti's avatar
Sebastien Moretti committed
245
246
print "\t   $IsExonerateHere\n";
undef $IsExonerateHere;
247
undef $VERSION;
248

Sebastien Moretti's avatar
Sebastien Moretti committed
249
for(my $m=0; $m<=$#original_names; $m++){
250
    printf("\t>%-3s  is  %s\n", $m, $original_names[$m]);
Sebastien Moretti's avatar
Sebastien Moretti committed
251
}
Sebastien Moretti's avatar
Sebastien Moretti committed
252
$date .= "_$$"; #Add PID to be more uniq
Sebastien Moretti's avatar
Sebastien Moretti committed
253
254
255
256



# run_name implementation for web server
257
$originalMSA = $run_name if ( $run_name ne '' );
Sebastien Moretti's avatar
Sebastien Moretti committed
258

259
260
unlink("$originalMSA.cds", "$originalMSA.cdsP", "$originalMSA.cdsP.html",
       "$originalMSA.out", "$originalMSA.boj",  "$originalMSA.log");
Sebastien Moretti's avatar
Sebastien Moretti committed
261
262


263
# Build the sequences, from the alignment, to perform blast search
264
open(my $LOG, '>', "$date.log") or die "\n\tCannot open LOG file\n\n";
265
EACH_SEQ:
266
267
268
269
270
271
272
273
for(my $r=0; $r<=$#original_names; $r++){
    my $fasta_header = $original_names[$r];
    my $short_name   = $original_names[$r];
    $fasta_header    =~ s{^>}{};               # Full FASTA header
    $short_name      =~ s{^([^\s\t]+).*$}{$1}; # Short name without description

    # Prepare sequence $r for BLAST
    open(my $READY2BLAST, '>', "$cache/$date.seq2blast.fas.$r");
274
    my $noGap = $original_seq[$r];
275
276
    $noGap    =~ s{\-}{}g; #remove gaps
    print {$READY2BLAST} ">$r\n$noGap\n";
277
    close $READY2BLAST;
Sebastien Moretti's avatar
Sebastien Moretti committed
278
279
280
    undef $noGap;


Sebastien Moretti's avatar
Sebastien Moretti committed
281
    my @equivalent_blast_hits; #Multiple BLAST hits with the same highest e-value/score
282
283

    if ( exists($template_NT->{$short_name}) && $template_NT->{$short_name} ne '' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
284
285
286
        # First try nucleotide template
        push @equivalent_blast_hits, 'My_own_seq';
        print "\n\n$r done\n\nNucleotide template for $r:\t$template_NT->{$short_name}\n\n*******************************************************************\n\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
287
    }
288
    elsif ( exists($template_AA->{$short_name}) && $template_AA->{$short_name} ne '' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
289
290
291
        # Second try amino acid/BLASTp template
        push @equivalent_blast_hits, $template_AA->{$short_name};
        print "\n\n$r done\n\nBlastP template for $r:\t$template_AA->{$short_name}\n\n*******************************************************************\n\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
292
293
    }
    else{
294
295
296
297
298
299
    # else send them to webblast
        if ($r==0){
            # Show BLAST parameters only for the first BLAST run
            display_blast_param();
        }
        run_BLAST("$cache/$date.seq2blast.fas.$r", $local, $blast_param->{'db1'}, $blast_exe, "$cache/$date.blastp.$r");
Sebastien Moretti's avatar
Sebastien Moretti committed
300

301
302
303

        # Get the BLASTp hit(s)  acc number
        open(my $BLASTP, '<', "$cache/$date.blastp.$r");
304
        while(<$BLASTP>){
305
            push @equivalent_blast_hits, $1  if ( $_ =~ /^>\s*\d+@?\w?\w?\w?__([^\s\|\.]+).*$/ ); #Warning: double '_'
306
307
        }
        close $BLASTP;
308
        unlink("$cache/$date.blastp.$r", "$cache/$date.seq2blast.fas.$r") if ( $tmp == 0 || exists($equivalent_blast_hits[0]) );
Sebastien Moretti's avatar
Sebastien Moretti committed
309
        if ( !exists($equivalent_blast_hits[0]) ){
310
            print "\n$r ...\tNo blast result found above thresholds for $fasta_header\n\n";
311
            buildFailureOutputFiles($r, 'No_BLASTp_Result', 'Not_searched', '');
312
313
314
315
            undef $original_seq[$r];
            undef $original_names[$r];
            next EACH_SEQ;
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
316
317
318
319
    }



320
    #When multiple equivalent blast hits  ==> use, and add, every hits
321
322
323
    my ($resultPOS, $resultBOJ) = ('', '');
    my @failureStatus = ('');
    my @intronStatus  = ('');
Sebastien Moretti's avatar
Sebastien Moretti committed
324
325
    for(my $qq=0; $qq<=$#equivalent_blast_hits; $qq++){
        print "\n$fasta_header  -->  [$equivalent_blast_hits[$qq]]\n";
326
327
328
329


        my @nt_GIs;
        my $intronStep = 0;
330
331
332
        if ( exists($template_NT->{$short_name}) && $template_NT->{$short_name} ne '' && $template_NT->{$short_name} !~ /^My_Seq$/i ){
            downloadSeq($cache, $date, '', '', $template_NT->{$short_name});
            @nt_GIs = $template_NT->{$short_name};
333
        }
334
        elsif ( exists($template_NT->{$short_name}) && $template_NT->{$short_name} ne '' && $template_NT->{$short_name} =~ /^My_Seq$/i ){
335
            open(my $MYSEQ, '>', "$cache/My_Seq-$$.fas");
336
            print {$MYSEQ} ">My_Seq-$$ for $short_name\n".$template_SEQ->{$short_name}."\n";
337
338
339
340
341
            close $MYSEQ;
            @nt_GIs = "My_Seq-$$";
        }
        else{
        #Prot ACC -> PUID
Sebastien Moretti's avatar
Sebastien Moretti committed
342
            my $protGI = blastPAcc2PGI($equivalent_blast_hits[$qq]);
343
            if ( $protGI eq '' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
344
345
                print "\tNo protein (prot GI) link found for $equivalent_blast_hits[$qq] in $fasta_header\n\n";
                @failureStatus = ('PUI_unavailable', $equivalent_blast_hits[$qq]);
346
347
                next;
            }
Sebastien Moretti's avatar
Sebastien Moretti committed
348
349
#           print "\t$protGI\n";

350
        #PUID -> nt GIs & GeneID
Sebastien Moretti's avatar
Sebastien Moretti committed
351
            my ($ntGIs, $geneID) = protGI2NTGIs($protGI, $equivalent_blast_hits[$qq]);
352
            if ( $ntGIs eq '' && $geneID eq '' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
353
354
                print "\tNo nucleotide (nt GIs) link found for $equivalent_blast_hits[$qq] in $fasta_header\n\n";
                @failureStatus = ('No_nt_link', $equivalent_blast_hits[$qq]);
355
356
357
358
359
360
361
                next;
            }
            print " linked with [$ntGIs $geneID]\n";


        #Get transcript and contig seq
            @nt_GIs = split(/,/, $ntGIs);
Sebastien Moretti's avatar
Sebastien Moretti committed
362
            downloadSeqFromGIs($cache, $date, '', '', @nt_GIs);
363
364
365
366
        #GeneID -> Chr
            my @geneIDs = split(/,/, $geneID);
            while(<@geneIDs>){
                my $geneID = $_;
Sebastien Moretti's avatar
Sebastien Moretti committed
367
368
                my ($chr, $amont, $aval) = geneID2Chr($geneID, $equivalent_blast_hits[$qq]);
                print "\n\tNo gene locus found for $equivalent_blast_hits[$qq] with $geneID in $fasta_header\n\n" if ($chr eq '' and $geneID ne '');
369
370
371

                if ( $chr ne '' ){
            #Get Chr seq
Sebastien Moretti's avatar
Sebastien Moretti committed
372
                    downloadSeq($cache, $date, $amont, $aval, $chr);
373
374
375
376
377
378
379
380
381
                    $intronStep = 1;

                    @nt_GIs = ("$chr:$amont-$aval", @nt_GIs);
                }
            }
        }


    #Align Nt seq with our protein query seq
Sebastien Moretti's avatar
Sebastien Moretti committed
382
        my ($POS, $BOJ) = runExonerate($exonerate_exe, $cache, $date, $r, $original_seq[$r], $fasta_header, $equivalent_blast_hits[$qq], $r, @nt_GIs);
383

Sebastien Moretti's avatar
Sebastien Moretti committed
384
        my ($POSresult, $bestNucleotide) = prepareResults4CDS($POS, $equivalent_blast_hits[$qq], $original_seq[$r], $fasta_header);
385
386
        $resultPOS .= $POSresult if ( $bestNucleotide ne '' && $POSresult ne '' && $resultPOS !~ /_G_$bestNucleotide/ );

387
388
        @intronStatus = ($POS->{0}, $equivalent_blast_hits[$qq], '')              if ( $intronStep==1 && exists($POS->{0}) && $POS->{0} ne '' && $POS->{0} =~ /\:/ );
        @intronStatus = ($POS->{0}, $equivalent_blast_hits[$qq], $bestNucleotide) if ( $intronStep==0 && ($bestNucleotide =~ /^N[CTWZG]_/ || $bestNucleotide =~ /^AC_/) );
389
390
391
392

        my ($intronLess, $nameLess) = ('', '');
        $intronLess = $POS->{0}       if ( %$BOJ eq 0 && exists($POS->{0}) && $POS->{0} ne '' );
        $nameLess   = $bestNucleotide if ( %$BOJ eq 0 && $intronStep==0 && ($bestNucleotide =~ /^N[CTWZG]_/ || $bestNucleotide =~ /^AC_/) );
Sebastien Moretti's avatar
Sebastien Moretti committed
393
        my ($BOJresult, $bestGenomic) = prepareResults4BOJ($BOJ, $equivalent_blast_hits[$qq], $original_seq[$r], $fasta_header, $intronLess, $nameLess);
394
        $resultBOJ .= $BOJresult      if ( $bestGenomic ne '' && $BOJresult ne '' && $resultBOJ !~ /_G_$bestGenomic[ \-]/ );
395
#    print " $intronStep $POS->{0} [@intronStatus] {$BOJresult} $bestNucleotide - $bestGenomic \t $BOJ->{0}\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
396
397
398
    }


399
    if ( $resultPOS eq '' ){
400
401
        buildFailureOutputFiles($r, $failureStatus[1],         $failureStatus[0],   '')  if ( $failureStatus[0] ne '' );
        buildFailureOutputFiles($r, $equivalent_blast_hits[0], 'Alignment_failure', '')  if ( $failureStatus[0] eq '' );
402
        next;
Sebastien Moretti's avatar
Sebastien Moretti committed
403
    }
404
405
    elsif ( $resultBOJ eq '' ){
        if ( $failureStatus[0] eq 'PUI_unavailable' || $failureStatus[0] eq 'No_nt_link' ){
406
            buildFailureBOJOutputFile($failureStatus[1], $failureStatus[0], $fasta_header, $original_seq[$r]);
407
408
        }
        elsif ( $failureStatus[0] eq '' && $intronStatus[0] ne '' ){
409
            buildIntronlessBOJOutputFile($intronStatus[1], $intronStatus[0], $intronStatus[2], $fasta_header, $original_seq[$r]);
410
411
        }
        else{
Sebastien Moretti's avatar
Sebastien Moretti committed
412
            buildFailureBOJOutputFile($equivalent_blast_hits[0], 'Unavailable', $fasta_header, $original_seq[$r]);
413
        }
414
        createCDSOutputFile($resultPOS, $original_seq[$r], $fasta_header);
Sebastien Moretti's avatar
Sebastien Moretti committed
415
416
    }
    else {
417
        createCDSOutputFile($resultPOS, $original_seq[$r], $fasta_header);
Sebastien Moretti's avatar
Sebastien Moretti committed
418
        createBOJOutputFile($resultBOJ, ${original_seq[$r]});
Sebastien Moretti's avatar
Sebastien Moretti committed
419
420
421
422
423
424
    }


    undef $original_seq[$r];
    undef $original_names[$r];
}
425
close $LOG;
426
unlink "$date.log" if ( -z "$date.log" );
Sebastien Moretti's avatar
Sebastien Moretti committed
427
428


Sebastien Moretti's avatar
Sebastien Moretti committed
429
430
checkAndCleanStderrFiles("$cache/${date}_ExonerateError") if ( -e "$cache/${date}_ExonerateError" );
checkOutputFiles();
Sebastien Moretti's avatar
Sebastien Moretti committed
431
unlink("$msa") if ( $converted==1 );
432
unlink 'blast_result.txt';
433
my @tmpFiles = glob($cache."/${date}*");
Sebastien Moretti's avatar
Sebastien Moretti committed
434
435
if ( exists($tmpFiles[0]) ){
    mkdir("$cache/${date}_TMP");
436
    move("$_", "$cache/${date}_TMP") while(<@tmpFiles>);
Sebastien Moretti's avatar
Sebastien Moretti committed
437
438
439
440
441
442
443
}


#Ouput Checker


#STDOUT Log output for our server
444
if ( -s "${originalMSA}.cds" || -s "${originalMSA}.out"){
Sebastien Moretti's avatar
Sebastien Moretti committed
445
446
    print "\n\nTERMINATION STATUS: SUCCESS\n";
    print "OUTPUT RESULTS\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
447
448
449
450
451
    print "    #### File Type=        MSA Format= fasta_CDS Name= ${originalMSA}.cds\n" 
        if ( -s "${originalMSA}.cds");
    print "    #### File Type=        MSA Format= fasta_CDS+Query Name= ${originalMSA}.cdsP\n" 
        if ( -s "${originalMSA}.cdsP");
    print "    #### File Type=        MSA Format= colored_fasta_CDS Name= ${originalMSA}.cdsP.html\n" 
Sebastien Moretti's avatar
Sebastien Moretti committed
452
        if ( Views::Html("${originalMSA}.cdsP") && -s "${originalMSA}.cdsP.html");
Sebastien Moretti's avatar
Sebastien Moretti committed
453
454
455
456
    print "    #### File Type=        MSA Format= Rejected_seq Name= ${originalMSA}.out\n" 
        if ( -s "${originalMSA}.out");
    print "    #### File Type=        MSA Format= fasta_Exon-Boundaries Name= ${originalMSA}.boj\n" 
        if ( -s "${originalMSA}.boj" && -s "${originalMSA}.cds" && $hideBOJ==0);
Sebastien Moretti's avatar
Sebastien Moretti committed
457
458
}
else{
Sebastien Moretti's avatar
Sebastien Moretti committed
459
    failure();
460
    exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
461
462
}

463
exit(0);
Sebastien Moretti's avatar
Sebastien Moretti committed
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507


=head1 NAME

ProtoGene.pl - Converts a peptidic alignment to a nucleotidic alignment through database search

=head1 SYNOPSIS

B<ProtoGene.pl> --msa=path_of_the_fasta_msa_file [options]

with these options available:

=over 4

=item I<--orgm>=targeted species in the database search

=item All_organisms [default], Bacteria, Viruses, Vertebrata, Eukaryota, Mammalia, Primates, Homo_sapiens, Gallus_gallus, Bos_taurus, Escherichia_coli, Arabidopsis_thaliana, Mus_musculus, Drosophila_Melanogaster, ...

=item I<--db>=targeted database in the database search

=item nr, pdb, swissprot, refseq_protein [default]

=item I<--local> to execute a local BLAST query with  --db=path_for_a_local_db_blast_formated

=item I<--template> to provide your own nucleotidic sequences following the cds file format

=item I<--revtrans> to reverse-translate sequences with no blast hit, in IUB (IUPAC) depiction code

=item They are removed from the alignement by default

=item I<--pep> to add the original peptide query beneath the back-translated sequence

=item I<--cache>=none, update, use, own_path_directory, old, empty

=item none: no cache usage, none temporary files are stored

=item update: use cache but update old files [default]

=item use: force use cache, whatever the age of files

=item own_path_directory: use my own directory, and its files

=item old: remove the old files in the cache directory

508
=item empty: empty completely the cache directory
Sebastien Moretti's avatar
Sebastien Moretti committed
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545

=item I<--version> to print version information

=item I<--help> to print this help message

=back

=head1 DESCRIPTION

PROTOGENE : Bona-Fide Back Translation of Protein Multiple Sequence Alignments
It converts a peptidic alignment to a nucleotidic alignment through database search.
Blast queries allow to get nucleotidic sequences from where peptidic sequences come from.
Exonerate aligns nucleotidic and peptidic sequences together.
PROTOGENE re-builds the original alignment with nucleotidic information it has gotten.

=head1 REQUIREMENT

=over 4

=item B<Perl 5.6 or better> is required !

=item Standard Perl modules B<lib>, B<strict>,  B<warnings>,  B<diagnostics>, B<Env> are required

=item and some other current ones : B<Getopt::Long>,  B<File::Which>,  B<File::Copy>, B<Mail::Send>

=item -

=item I<exonerate> from http://www.ebi.ac.uk/~guy/exonerate/

=item I<blast> from http://www.ncbi.nlm.nih.gov/BLAST/download.shtml or http://blast.wustl.edu/

=back

=head1 VERSION

=over 8

546
=item version 4.0.0
Sebastien Moretti's avatar
Sebastien Moretti committed
547

548
=item on Apr 02nd, 2011
Sebastien Moretti's avatar
Sebastien Moretti committed
549
550
551
552
553
554
555
556
557

=back

=head1 AUTHORS

=over 8

=item Sebastien MORETTI

Sebastien Moretti's avatar
Sebastien Moretti committed
558
=item moretti.sebastien [AT] gmail.com
Sebastien Moretti's avatar
Sebastien Moretti committed
559

560
=item Vital-IT computing center
Sebastien Moretti's avatar
Sebastien Moretti committed
561

562
=item Swiss Institute of Bioinformatics
Sebastien Moretti's avatar
Sebastien Moretti committed
563

564
=item Lausanne, Switzerland
Sebastien Moretti's avatar
Sebastien Moretti committed
565

566
=item http://www.vital-it.ch/
Sebastien Moretti's avatar
Sebastien Moretti committed
567
568
569
570
571
572
573
574

=back

=cut


######################  Management  ######################

575
# Failure declaration
Sebastien Moretti's avatar
Sebastien Moretti committed
576
sub failure {
577
578
579
    print {*STDERR} "\n\nFATAL\n\nTERMINATION STATUS: FAILURE\n\n";
    return;
}
Sebastien Moretti's avatar
Sebastien Moretti committed
580

Sebastien Moretti's avatar
Sebastien Moretti committed
581
# Check external programs presence in the PATH
582
sub checkExternalSoftware {
Sebastien Moretti's avatar
Sebastien Moretti committed
583
584
585
586
587
588
    for my $exe ( @_ ){
        if ( ! which($exe) ){
            failure();
            print {*STDERR} "\t'$exe' program is not reachable\n\tIt could not be in your PATH or not installed\n\n";
            exit(1);
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
589
    }
Sebastien Moretti's avatar
Sebastien Moretti committed
590
    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
591
592
}

593
594
# Check cache directory accessibility: Test and change rights, if required, of the cache directory
sub checkCacheAccessibility {
595
    my ($cacheDir) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
596

597
    if ( -d "$cacheDir/" && -w "$cacheDir/" && -x "$cacheDir/" && -r "$cacheDir/" ){
598
        # Good
Sebastien Moretti's avatar
Sebastien Moretti committed
599
    }
600
601
602
    elsif ( -d "$cacheDir/" ){
        if ( -o "$cacheDir/" ){
            chmod 0754, "$cacheDir/";
603
        }
604
        else{
605
606
607
            failure();
            print {*STDERR} "\tPermissions do NOT seem to be valid for the '$cacheDir' cache directory\n\n";
            exit(2);
608
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
609
610
    }
    else{
611
        mkdir "$cacheDir/"; # FIXME: mkdir is not recursive !
612
        if ( -o "$cacheDir/" ){
613
            chmod 0754, "$cacheDir/"; #0754 is in octal, so 754 or '0754' are not good !
614
615
        }
        else{
616
617
618
            failure();
            print {*STDERR} "\tPermissions do NOT seem to be valid for the '$cacheDir' cache directory\n\n";
            exit(2);
619
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
620
    }
621

622
    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
623
624
}

625
# Cache Behavior Management
Sebastien Moretti's avatar
Sebastien Moretti committed
626
sub cacheManagement{
627
    my ($cache, $cacheDir, $inScript) = @_;
628

629
    if ( $cache eq 'empty' ){
630
631
        unlink <$cacheDir/*.fas>;
        unlink <$cacheDir/*_ExonerateError>;
632
633
        unlink("$cacheDir/webblast.log", "$cacheDir/web_tempo.result", "$cacheDir/debug.tempo");
        print "\n\tcache directory [$cacheDir] is empty\n\n";
634
        exit 0;
Sebastien Moretti's avatar
Sebastien Moretti committed
635
    }
636
    elsif ( $cache eq 'old' ){
637
638
639
640
641
        unlink    grep { -M $_ > $cacheStorageTime }    glob($cacheDir.'/*.fas'); # Grep list only with $cacheStorageTime days old files

        if ( $inScript==0 ){
            print "\n\tcache directory [$cacheDir] is empty of its old files\n\n";
            exit 0;
642
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
643
    }
644
    elsif ( $cache eq 'update' || $cache eq 'use' ){
645
646
        # DEFAULT: Limit cached files to less than $cacheStorageTime days old ones for 'update'
        $cache = $cacheDir;
Sebastien Moretti's avatar
Sebastien Moretti committed
647
    }
648
    elsif ( $cache eq 'none' ){
649
        # Do not use cache
Sebastien Moretti's avatar
Sebastien Moretti committed
650
    }
651
652
653
654
    elsif ( -d $cache ){
        die "\n\tProtoGene cannot access to your directory\n\n"  if ( ! -x $cache );
        die "\n\tProtoGene cannot write into your directory\n\n" if ( ! -w $cache );
        die "\n\tProtoGene cannot read into your directory\n\n"  if ( ! -r $cache );
Sebastien Moretti's avatar
Sebastien Moretti committed
655
656
    }
    else {
657
658
659
        failure();
        print {*STDERR} "\n\tWrong cache argument, or your cache folder '$cacheDir' doesn't seem to exist\n\n";
        exit(3);
Sebastien Moretti's avatar
Sebastien Moretti committed
660
661
662
663
664
665
666
667
668
669
    }

    return($cache);
}

##########################################################


######################    Fasta    ######################

Sebastien Moretti's avatar
Sebastien Moretti committed
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
# Check input aln format
sub isFastaFile{
    my ($msa, $converted) = @_;

    if ( -s "$msa" ){ # Check is the file exists and is not empty
        # Is FASTA format ?
        my $firstLine = `head $msa`;
        return($msa, 0) if ( $firstLine =~ /^>/ );

        # Convert no fasta format to fasta with readseq
        my $reformat = which('readseq') || '';
        if ( $reformat ne '' ){
            system("$reformat -a -f8 $msa > $msa.007")==0 or do {failure(); print {*STDERR} "\n\tEnable to convert your file to fasta format\n\n";};
            $msa      .= '.007';
            $converted = 1;
685
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
686
687
    }
    else {
Sebastien Moretti's avatar
Sebastien Moretti committed
688
        failure();
689
690
        print {*STDERR} "\tThe file \'$msa\' doesn't not seem to be reachable\n\n";
        exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
691
692
    }

Sebastien Moretti's avatar
Sebastien Moretti committed
693
    return($msa, $converted);
Sebastien Moretti's avatar
Sebastien Moretti committed
694
695
696
697
}

#Prepare Fasta Header of the query name to incorporate our annotations
sub fastaHeaders4ProtoGene{
698
    my ($QueryName) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
699
700

    $QueryName =~ s/  +/ /g;
701
702
703
704
705
706
707
708
709
710
    my ($acc, $desc) = ($QueryName, $QueryName);
    $acc  =~ s/^ *([^ ]*) .*$/$1/;
    $acc  =~ s/\s+$//g;
    $acc  =  $acc.'_G_@@';

    $desc =~ s/^ *[^ ]* *(.*)$/$1/;
    $desc =~ s/^[\| \.]+//;
    $desc =~ s/[\|\. ]+$//;

    return($acc.$desc);
Sebastien Moretti's avatar
Sebastien Moretti committed
711
712
713
714
}

#Prepare Fasta Header to put it on the fasta header of the result outputs
sub getFastaHeaderAnnot{
715
    my ($bestTarget) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
716
717

    #Get annotation from nucleotide file header
718
719
720
    open(my $BEST, '<', "$cache/${bestTarget}.fas");
    my $annot=<$BEST> || '';
    close $BEST;
Sebastien Moretti's avatar
Sebastien Moretti committed
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
    chomp($annot);
    $annot =~ s/^ *[^ ]* //;
    $annot =~ s/[\. ]*$//;
    $annot =~ s/^ *//;
    $annot =~ s/  +/ /g;

    return($annot);
}

#########################################################


###################### Translation ######################

#Direct Reverse translation
sub reverse_trad{
737
    my ($aa) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
738

739
740
741
742
743
744
    my %reverse_code = ('A' => 'GCN', 'F' => 'TTy', 'K' => 'AAr', 'P' => 'CCN', 'T' => 'ACN',
                        'C' => 'TGy', 'G' => 'GGN', 'L' => 'yTN', 'Q' => 'CAr', 'V' => 'GTN',
                        'D' => 'GAy', 'H' => 'CAy', 'M' => 'ATG', 'R' => 'mGN', 'W' => 'TGG',
                        'E' => 'GAr', 'I' => 'ATh', 'N' => 'AAy', 'S' => 'wsN', 'Y' => 'TAy',
                        'X' => 'NNN', '-' => '---', '*' => 'Trr',
                       ); # '*' is for stop codon
Sebastien Moretti's avatar
Sebastien Moretti committed
745

746
    my $triplet = 'nnn';
747
    while( my ($x, $y) =each(%reverse_code) ){
748
749
750
751
        if ( uc($aa) eq $x ){
            $triplet = $y;
            last;
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
752
753
754
755
756
757
758
759
760
    }

    return($triplet);
}

#########################################################


####################### webblast #######################
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
sub display_blast_param {
    print "
             Program :                   BLASTp
             Evalue threshold :          $blast_param->{'evalue'}
             Matrix :                    $blast_param->{'matrix'}
             Filter :                    $blast_param->{'filter'}
             Equivalent best hits used : $blast_param->{'nbesthits'}

             Database:                   NCBI $blast_param->{'db1'}
             Blast_identity_threshold :  $blast_param->{'identity1'}
             Cover threshold :           $blast_param->{'cover1'}
          if no match
             Database:                   NCBI $blast_param->{'db2'}
             Blast_identity_threshold :   $blast_param->{'identity2'}
             Cover threshold :            $blast_param->{'cover2'}
\n";
    return;
}

sub run_BLAST{
    my ($infile, $whatBlast, $db, $blast_exe, $outfile) = @_;

    my $blast_at = $whatBlast==2  ? '-gigablast=yes'
                 : $whatBlast==1  ? "-blast_dir=$blast_exe"
                 :                  '';

    # BLAST run 1
    system("$webblast_exe -program=blastp -database=$db $blast_at -infile=$infile -matrix=$blast_param->{'matrix'} -evalue=$blast_param->{'evalue'} -method=geneid -filter=$blast_param->{'filter'} -organism=$species -identity=$blast_param->{'identity1'} -cover=$blast_param->{'cover1'} -hits=$blast_param->{'nbesthits'} -quiet=on -outfile=$outfile");


    if ( -z "$outfile" ){
    # BLAST run 2 if no hit with 100% identity and coverage: refseq100 -> nr95  OR  nr100 -> nr95
        print "run BLAST.. against $blast_param->{'db2'} because there was no acceptable hit against $blast_param->{'db1'}"
794
            if ( $whatBlast != 1 && $db ne 'nr' );
795
        print "run BLAST.. again, with decreased thresholds, because there was no acceptable hit against $db"
796
            if ( $whatBlast == 1 || $db eq 'nr' );
797
798

        system("$webblast_exe -program=blastp -database=$blast_param->{'db2'} $blast_at -infile=$infile -matrix=$blast_param->{'matrix'} -evalue=$blast_param->{'evalue'} -method=geneid -filter=$blast_param->{'filter'} -organism=$species -identity=$blast_param->{'identity2'} -cover=$blast_param->{'cover2'} -hits=$blast_param->{'nbesthits'} -quiet=on -outfile=$outfile");
Sebastien Moretti's avatar
Sebastien Moretti committed
799
800
    }

801
802
803
804
805
806
807
    move('webblast.log',     "$cache"); #Move temporary webblast.pl files
    move('web_tempo.result', "$cache");
    move('debug.tempo',      "$cache");

    unlink("$cache/webblast.log")     if ( -z "$cache/webblast.log" );
    unlink("$cache/web_tempo.result") if ( -z "$cache/web_tempo.result" );
    unlink("$cache/debug.tempo")      if ( -z "$cache/debug.tempo" );
Sebastien Moretti's avatar
Sebastien Moretti committed
808

809
    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
810
811
812
813
814
815
816
817
818
819
}


########################################################


##################### NCBI requests #####################

#Prot ACC -> PUID
sub blastPAcc2PGI{
820
    my ($blastHit) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
821

822
823
824
    my $protGI = '';
    my $count  = 0;
    GET_PUI:
Sebastien Moretti's avatar
Sebastien Moretti committed
825
    for(my $rep=0;$rep <= 4; $rep++){
826
        $count++;
827
828
#        system("wget -q -O $cache/${date}_${blastHit}protGi.tmp 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?db=protein&term=${blastHit}[pacc]&retmode=xml&tool=ProtoGene&email=smoretti\@unil.ch'");
        system("wget -q -O $cache/${date}_${blastHit}protGi.tmp 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?db=protein&term=${blastHit}&retmode=xml&tool=ProtoGene&email=smoretti\@unil.ch'");
829
830
831
        open(my $GIP, '<', "$cache/${date}_${blastHit}protGi.tmp");
        PROT_GI:
        while(<$GIP>){
832
            if ( $_ =~ /^.*<Id>(\d+)<\/Id>.*$/ ){
833
834
835
836
837
838
839
                $protGI = $1;
                last PROT_GI; #OK if only one PUID
            }
        }
        close $GIP;
        $rep = $rep-15 if ( -z "$cache/${date}_${blastHit}protGi.tmp" && $count==1 );
        last GET_PUI if ( $protGI =~ /^\d+$/ );
Sebastien Moretti's avatar
Sebastien Moretti committed
840
    }
841
    unlink("$cache/${date}_${blastHit}protGi.tmp") if ( $tmp == 0 || $protGI ne '' );
Sebastien Moretti's avatar
Sebastien Moretti committed
842
843
844
845
846
847

    return($protGI);
}

#PUID -> nt GIs
sub protGI2NTGIs{
848
    my ($protGI, $blastHit) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
849

850
851
852
853
    my $ntGIs  = '';
    my $geneID = '';
    my $count  = 0;
    GET_NTUI:
Sebastien Moretti's avatar
Sebastien Moretti committed
854
    for(my $rep=0;$rep <= 4; $rep++){
855
        $count++;
Sebastien Moretti's avatar
Sebastien Moretti committed
856
        #nt GIs
857
        system("wget -q -O $cache/${date}_${blastHit}nucleo.tmp 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=protein&db=nuccore,nucleotide,gene&id=$protGI&retmode=xml&tool=ProtoGene&email=smoretti\@unil.ch'");
858
859
860
861
862
863
864
865
866
867
868
        open(my $GIN, '<', "$cache/${date}_${blastHit}nucleo.tmp");
        my $flag = 0;
        NT_GI:
        while(<$GIN>){
            if ( $_   =~ /<LinkName>protein_nuc[a-z]+<\/LinkName>/ ){ #for nuccleotide and nuccore
                $flag = 1;
            }
            elsif ( $_ =~ /\<LinkName\>protein_gene\<\/LinkName\>/ ){
                $flag  = 2;
            }
            elsif ( $flag==1 && $_ =~ /^.*\<Id\>(\d+)\<\/Id\>.*$/ ){
869
870
871
                my $match = $1;
                $ntGIs  .= ",$match" if ( $ntGIs  ne '' && $ntGIs  !~ /,$match,/ && $ntGIs  !~ /^$match,/ );
                $ntGIs   = $match    if ( $ntGIs  eq '' );
872
873
            }
            elsif ( $flag==2 && $_ =~ /^.*\<Id\>(\d+)\<\/Id\>.*$/ ){
874
875
876
                my $match = $1;
                $geneID .= ",$match" if ( $geneID ne '' && $geneID !~ /,$match,/ && $geneID !~ /^$match,/ );
                $geneID  = $match    if ( $geneID eq '' );
877
878
879
880
881
            }
        }
        close $GIN;
        $rep = $rep-15 if ( -z "$cache/${date}_${blastHit}nucleo.tmp" && $count==1 );
        last GET_NTUI if ( $ntGIs ne '' || $geneID ne '' );
Sebastien Moretti's avatar
Sebastien Moretti committed
882
    }
883
    unlink("$cache/${date}_${blastHit}nucleo.tmp") if ( $tmp==0 || ($ntGIs ne '' || $geneID ne '') );
Sebastien Moretti's avatar
Sebastien Moretti committed
884

885
    return($ntGIs, $geneID);
Sebastien Moretti's avatar
Sebastien Moretti committed
886
887
888
}

sub geneID2Chr{
889
    my ($geneID, $blastHit) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
890

891
892
893
894
    my $chr   = '';
    my ($amont, $aval) = ('', '');
    my $count = 0;
    GET_CHR:
Sebastien Moretti's avatar
Sebastien Moretti committed
895
    for(my $rep=0;$rep <= 8; $rep++){
896
        $count++;
Sebastien Moretti's avatar
Sebastien Moretti committed
897
        #Gene Acc
898
899
#        system("wget -q -O $cache/${date}_${blastHit}gene.tmp 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=gene&id=$geneID&retmode=xml&tool=ProtoGene&email=smoretti\@unil.ch'");
        system("wget -q -O $cache/${date}_${blastHit}gene.tmp 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esummary.fcgi?db=gene&id=$geneID&retmode=xml&tool=ProtoGene&email=smoretti\@unil.ch'");
900
901
902
903
904
905
906
        open(my $GACC, '<', "$cache/${date}_${blastHit}gene.tmp");
        my $flag = 0;
        GENE_CHR:
        while(<$GACC>){
            if ( $_  =~ /\<ERROR\>Empty id list \- nothing todo\<\/ERROR\>/ && $flag==0 ){
                $rep = $rep-15 if ( $count==1 );
            }
907
908
909
#            if ( $_   =~ /\<Entrezgene_locus\>/ && $flag==0 ){
            if ( $_   =~ /<Item Name="GenomicInfoType" Type="Structure">/ && $flag==0 ){
               $flag = 1;
910
            }
911
912
913
914
915
916
917
#            elsif ( $_ =~ /\<Gene-commentary_type value=\"genomic\"\>/ && $flag==1 ){
#                $flag  = 2;
#            }
#            elsif ( $_ =~ /\<Gene-commentary_type value=/ && $flag==2 ){
#                last GENE_CHR;
#            }
#            elsif ( $_ =~ /\<Gene-commentary_accession\>([\w\_\-\.]+)\<\/Gene-commentary_accession\>/ && $flag==2 ){
Sebastien Moretti's avatar
Sebastien Moretti committed
918
            elsif ( $_ =~ /<Item Name="ChrAccVer" Type="String">([^<]+?)\.?\d*<\/Item>/ && $flag==1 ){
919
920
921
                $chr   = $1;
                $flag  = 3;
            }
922
923
#            elsif ( $_ =~ /\<Seq-interval_from\>(\d+)\<\/Seq-interval_from\>/ && $flag==3 ){
            elsif ( $_ =~ /<Item Name="ChrStart" Type="Integer">(\d+)<\/Item>/ && $flag==3 ){
924
925
926
                $amont = $1;
                $flag  = 4;
            }
927
928
#            elsif ( $_ =~ /\<Seq-interval_to\>(\d+)\<\/Seq-interval_to\>/ && $flag==4 ){
            elsif ( $_ =~ /<Item Name="ChrStop" Type="Integer">(\d+)<\/Item>/ && $flag==4 ){
929
930
931
932
933
934
935
                $aval  = $1;
                last GENE_CHR;
            }
        }
        close $GACC;
        $rep = $rep-15 if ( -z "$cache/${date}_${blastHit}gene.tmp" && $count==1 );
        last GET_CHR   if ( $chr ne '' && $amont =~ /^\d+$/ && $aval =~ /^\d+$/ );
Sebastien Moretti's avatar
Sebastien Moretti committed
936
    }
937
938
939
940
    unlink("$cache/${date}_${blastHit}gene.tmp") if ( $tmp == 0 || $chr ne '' );
    if ( $amont eq '' || $aval eq '' ){
        $amont = '';
        $aval = '';
Sebastien Moretti's avatar
Sebastien Moretti committed
941
    }
942
943
944
945
946
947
948
949
950
    elsif ( $amont ne '' && $aval ne '' ){
        if ( $amont > $aval ){
            $amont = $amont+5000;
            $aval  = $aval-5000;
        }
        else{
            $amont = $amont-5000;
            $aval  = $aval+5000;
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
951
    }
952
953
    $amont = 1 if ( $amont ne '' && $amont <= 0 );
    $aval  = 1 if ( $aval ne ''  && $aval <= 0 );
Sebastien Moretti's avatar
Sebastien Moretti committed
954

955
    return($chr, $amont, $aval);
Sebastien Moretti's avatar
Sebastien Moretti committed
956
957
958
}

sub downloadSeqFromGIs{
959
    my ($cache, $date, $amont, $aval, @acc) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
960

961
    GET_SEQ:
Sebastien Moretti's avatar
Sebastien Moretti committed
962
    for(my $a=0; $a<=$#acc; $a++){
963
964
965
966
        my $whatNumber = 0;
        if ( $amont =~ /^\d+$/ && $aval =~ /^\d+$/ ){
            GET_CHR_SEQ:
            for(my $rep=0;$rep <= 4; $rep++){
967
                system("wget -q -O $cache/${acc[$a]}-${amont}.fas 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nucleotide&id=${acc[$a]}&rettype=fasta&retmode=text&from=$amont&to=$aval&tool=ProtoGene&email=smoretti\@unil.ch'") if ( !-e "$cache/${acc[$a]}-${amont}.fas" || -z "$cache/${acc[$a]}-${amont}.fas" || $cache eq 'none' || ($cache eq 'update' && -M "$cache/${acc[$a]}-${amont}.fas" > $cacheStorageTime ) );
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
                open(my $CIBLE, '<', "$cache/${acc[$a]}-${amont}.fas");
                my $counter = 0;
                my $lines   = 0;
                CHECK_CHR_SEQ:
                while(<$CIBLE>){
                    $lines++              if ( $_ !~ /^>/ );
                    $counter = $counter+2 if ( $counter==1 && $_ !~ /^\w/ && $lines==1 && $_ !~ /^>/ );
                    $counter++            if ( $_ =~ /^>/ );
                    $counter = $counter+2 if ( $_ !~ /^>/ && ($_ =~ /Error:/ || $_ =~ /[<>]/) );
#                   last CHECK_CHR_SEQ    if ( $_ !~ /^>/ && $counter==1 && $_ =~ /^\w/ );
                }
                close $CIBLE;
                $whatNumber++;
                last GET_CHR_SEQ if ( $whatNumber==20 );
                if ( $counter != 1 ){
                    $rep = $rep-1;
                    unlink("$cache/${acc[$a]}-${amont}.fas");
                }
            }
        }
        else{
            GET_OTHER_SEQ:
            for(my $rep=0;$rep <= 4; $rep++){
991
                system("wget -q -O $cache/${acc[$a]}.fas 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nucleotide&id=${acc[$a]}&rettype=fasta&retmode=text&tool=ProtoGene&email=smoretti\@unil.ch'") if ( !-e "$cache/${acc[$a]}.fas" || -z "$cache/${acc[$a]}.fas" || $cache eq 'none' || ($cache eq 'update' && -M "$cache/${acc[$a]}.fas" > $cacheStorageTime ) );
992
993
994
995
996
997
998
999
1000
                open(my $CIBLE, '<', "$cache/${acc[$a]}.fas");
                my $counter = 0;
                my $lines   = 0;
                CHECK_OTHER_SEQ:
                while(<$CIBLE>){
                    $lines++              if ( $_ !~ /^>/ );
                    $counter = $counter+2 if ( $counter==1 && $_ !~ /^\w/ && $lines==1 && $_ !~ /^>/ );
                    $counter++            if ( $_ =~ /^>/ );
                    $counter = $counter+2 if ( $_ !~ /^>/ && ($_ =~ /Error:/ || $_ =~ /[<>]/) );