ProtoGene.pl 62 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
18
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
19

Sebastien Moretti's avatar
Sebastien Moretti committed
20
use File::Copy qw(move);                   # Avoid external 'mv' command usage
21
use LWP::Simple;                           # To test gigablaster availability
Sebastien Moretti's avatar
Sebastien Moretti committed
22
23
24

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

25
26
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
27
use Exonerate;                             # Exonerate runner, parser, ...
28
use Views;                                 # Non-text outputs, e.g. HTML/CSS
29
#use CheckOutput;                         # Check output for cds consistancy with query
Sebastien Moretti's avatar
Sebastien Moretti committed
30
31
32



33
################## CONFIGURATION ##################
Sebastien Moretti's avatar
Sebastien Moretti committed
34
$ENV{'PATH'}        .= ':/mnt/local/bin/:./';                               # Additional path for executables
35
36

my $cachePath        = '/scratch/cluster/monthly/t_coffee/ProtoGene_Cache'; # Cache directory
37
$cachePath = '/Users/smoretti/Documents/ProtoGene/TTMMPP'; #FIXME
38
39
40
41
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
###################################################
Sebastien Moretti's avatar
Sebastien Moretti committed
42
43


44
my $VERSION  = '4.0.0';
Sebastien Moretti's avatar
Sebastien Moretti committed
45

Sebastien Moretti's avatar
Sebastien Moretti committed
46
my $webblast_exe  = 'webblast.pl';
47
my $blast_exe     = 'blastall';            # Or wu-blastall for Wu-BLAST; for local blast usage
Sebastien Moretti's avatar
Sebastien Moretti committed
48
my $exonerate_exe = 'exonerate-1.0';       # Exonerate 1.0 because current parser only works with this version
49
50
51
52
53
54


################## Option management
my ($msa, $revtrans, $pep, $hideBOJ, $run_name, $template, $lim, $cache) = ('', 0, 0, 0, '', '', 0, 'update');
my ($debug, $tmp)                                                        = (0, 0);
my ($db, $species, $local, $giga)                                        = ('refseq_protein', 'All_organisms', 0, 0);
Sebastien Moretti's avatar
Sebastien Moretti committed
55
56
57
58
59
60
61
62
63
64
65
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
66
            'local'           => \$local,      # Use to specify a local db query, defined in $local_db
Sebastien Moretti's avatar
Sebastien Moretti committed
67
68
69
70
71
72
73
74
75
76
            '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
77
           );
Sebastien Moretti's avatar
Sebastien Moretti committed
78
79
my $test_option_values = Getopt::Long::GetOptions(%opts);

80
81

################## Short help message
82
83
if ( !$test_option_values || ($msa eq '' && $cache ne 'empty' && $cache ne 'old') ){
#|| $species eq '' || $db eq '' ){
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
    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, ...
\t                default is 'All_organisms'
\t         --db=nr, pdb, swissprot, refseq_protein
\t                default is 'refseq_protein'
\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";
109
    exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
110
}
111
112


113
114
################## Check cache directory accessibility
checkCacheAccessibility($cachePath);
115

116
117
# Cache management
$cache = cacheManagement($cache, $cachePath, 0) || '.';
118

119
120
# Remove old files from the cache directory everytime ProtoGene is running
cacheManagement('old', $cachePath, 1);
121
122


123
124

################## Check external software in the PATH
Sebastien Moretti's avatar
Sebastien Moretti committed
125
126
#TODO: remove that for production
#checkExternalSoftware($webblast_exe, $exonerate_exe);
127
128
if ( $local ){
    checkExternalSoftware($blast_exe);
Sebastien Moretti's avatar
Sebastien Moretti committed
129
130
131
132
}



133
134
135
136
137
################## Check input files
# Open and check template file
my ($template_NT, $template_AA, $template_SEQ);
($template_NT, $template_AA, $template_SEQ) = checkTemplate($template) if ( $template );

Sebastien Moretti's avatar
Sebastien Moretti committed
138

Sebastien Moretti's avatar
Sebastien Moretti committed
139
140
141
142
143
# 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
144

145
my $fasta_checker = -1;
146
my (@input_order, @original_names, @original_seq); #Use 3 lists in parallel : original_names, seq, order(1->n)
Sebastien Moretti's avatar
Sebastien Moretti committed
147
148
open( my $MSA, '<', "$msa" ) or die "\n\tCannot open your MSA file\n\n";
READ_MSA:
149
while(<$MSA>){
Sebastien Moretti's avatar
Sebastien Moretti committed
150
151
    next READ_MSA if ( $_ =~ /^#/ );

152
    if ( $_ =~ /^>/ ){
153
154
        $fasta_checker++;
        my $name = $_;
Sebastien Moretti's avatar
Sebastien Moretti committed
155
156
157
        $name    =~ s{\r\n}{}g; #Remove return lines from windows OS '^M'
        $name    =~ s{\r}{\n}g;
        $name    =~ s{^>[ \t]+}{>};
158
        chomp($name);
Sebastien Moretti's avatar
Sebastien Moretti committed
159
160
161

        push @original_names, $name;
        push @input_order,    $fasta_checker;
Sebastien Moretti's avatar
Sebastien Moretti committed
162
    }
Sebastien Moretti's avatar
Sebastien Moretti committed
163
164
165
166
167
168
169
170
171
172
    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
173
        #fasta sequence on 1 line
174
        if ( !exists($original_seq[$fasta_checker]) ){
Sebastien Moretti's avatar
Sebastien Moretti committed
175
            push @original_seq, $sequence;
176
177
        }
        else {
Sebastien Moretti's avatar
Sebastien Moretti committed
178
            $original_seq[$fasta_checker] .= $sequence;
179
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
180
181
    }
}
182
close $MSA;
Sebastien Moretti's avatar
Sebastien Moretti committed
183
184

#Are there FASTA sequences ?
185
if ( $fasta_checker == -1 ){
Sebastien Moretti's avatar
Sebastien Moretti committed
186
    failure();
Sebastien Moretti's avatar
Sebastien Moretti committed
187
188
    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";
189
    exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
190
}
191
elsif ( $lim >0 && $fasta_checker > $lim ){
Sebastien Moretti's avatar
Sebastien Moretti committed
192
    failure();
193
    print {*STDERR} "\tThe FASTA file is too large, try with less than $lim sequences\n\tor split your file\n\n";
194
195
    exit(1);
}
196
elsif ( exists( $original_seq[0] ) && $original_seq[0] =~ /[acgtu]/i ){
197
    my $first_seq = $original_seq[0];
Sebastien Moretti's avatar
Sebastien Moretti committed
198
199
200
201
202
203
    #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;
204
    if ( ($a+$c+$g+$t) >= (($a+$c+$g+$t+$non)*80/100) ){
Sebastien Moretti's avatar
Sebastien Moretti committed
205
        failure();
Sebastien Moretti's avatar
Sebastien Moretti committed
206
207
        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";
208
209
        exit(1);
    }
Sebastien Moretti's avatar
Sebastien Moretti committed
210
211
212
213
}
undef $fasta_checker;


Sebastien Moretti's avatar
Sebastien Moretti committed
214
# Check GigaBlaster status if used
215
216
if ( $giga==1 ){
    if ( head('http://www.igs.cnrs-mrs.fr/Giga2/~database/remoteblast.cgi') ){
Sebastien Moretti's avatar
Sebastien Moretti committed
217
        $local = 2;
218
219
    }
    else{
220
        $local = 0;
221
222
223
    }
}

Sebastien Moretti's avatar
Sebastien Moretti committed
224

Sebastien Moretti's avatar
Sebastien Moretti committed
225
226
227
228
# Temporary file extension
my $date = sprintf("%d-%02d-%02d_%02d%02d", localtime->year() + 1900, localtime->mon(), localtime->mday(), localtime->hour, localtime->min);


229

Sebastien Moretti's avatar
Sebastien Moretti committed
230
#Start main program with version # of programs and list original queries
231
print "\n\t   Protogene\t$VERSION\t$date\n\n";
232
open(my $EXONERATEISHERE, "$exonerate_exe --version |");
233
my $IsExonerateHere = <$EXONERATEISHERE>;
234
close $EXONERATEISHERE;
Sebastien Moretti's avatar
Sebastien Moretti committed
235
236
print "\t   $IsExonerateHere\n";
undef $IsExonerateHere;
237
undef $VERSION;
Sebastien Moretti's avatar
Sebastien Moretti committed
238
239
240
for(my $m=0; $m<=$#original_names; $m++){
    print "\t$original_names[$m]\n";
}
Sebastien Moretti's avatar
Sebastien Moretti committed
241
$date .= "_$$"; #Add PID to be more uniq
Sebastien Moretti's avatar
Sebastien Moretti committed
242
243
244
245



# run_name implementation for web server
246
$originalMSA = $run_name if ( $run_name ne '' );
Sebastien Moretti's avatar
Sebastien Moretti committed
247
248
249

unlink("${originalMSA}.cds");
unlink("${originalMSA}.cdsP");
Sebastien Moretti's avatar
Sebastien Moretti committed
250
unlink("${originalMSA}.cdsP.html");
Sebastien Moretti's avatar
Sebastien Moretti committed
251
252
253
254
255
256
unlink("${originalMSA}.out");
unlink("${originalMSA}.boj");



##Build the sequences, from the alignment, to perform webblast
257
258
my $quiet = '';
EACH_SEQ:
Sebastien Moretti's avatar
Sebastien Moretti committed
259
for(my $r=0; $r<=$#input_order; $r++){
260
261
262
263
    my $right_name = $original_names[$r];
    my $templ_name = $original_names[$r];
    $right_name    =~ s/^>//;
    $templ_name    =~ s/^([^\s\t]+).*$/$1/;
Sebastien Moretti's avatar
Sebastien Moretti committed
264

265
    open(my $READY2BLAST, '>', "$cache/${date}_seq2blast.fas${r}");
266
267
    my $noGap = $original_seq[$r];
    $noGap    =~ s/[\.-]//g; #remove gaps
268
269
    print {$READY2BLAST} ">$input_order[$r]\n$noGap\n";
    close $READY2BLAST;
Sebastien Moretti's avatar
Sebastien Moretti committed
270
271
272
273
    undef $noGap;


    my @multipleequalblasthits;
274
    if ( exists($template_NT->{$templ_name}) && $template_NT->{$templ_name} ne '' ){
275
        @multipleequalblasthits = (@multipleequalblasthits, 'My_own_seq'); #To change for ?
276
        print "\n\n$r done\n\nNucleotidic template for $r:\t$template_NT->{$templ_name}\n\n**********************************************************************\n\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
277
    }
278
279
280
    elsif ( exists($template_AA->{$templ_name}) && $template_AA->{$templ_name} ne '' ){
        @multipleequalblasthits = (@multipleequalblasthits, $template_AA->{$templ_name});
        print "\n\n$r done\n\nBlastP template for $r:\t$template_AA->{$templ_name}\n\n**********************************************************************\n\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
281
282
    }
    else{
283
    ##   --> send them to webblast
284
        $quiet = '-quiet=on ' if ( $r>0 ); #Show webblast parameters only for the first webblast run
285
        launchWebblast("$cache/${date}_seq2blast.fas${r}", $quiet, $local, $db, $species, $blast_exe, "$cache/${date}.blastp${r}");
Sebastien Moretti's avatar
Sebastien Moretti committed
286

287
    ##Get the BLASTp hit(s)  acc number
288
289
290
291
292
293
294
295
        open(my $BLASTP, '<', "${cache}/${date}.blastp${r}");
        while(<$BLASTP>){
            @multipleequalblasthits = (@multipleequalblasthits, $1) if ( $_ =~ /^>\s*\d+@?\w?\w?\w?__([^\s\|\.]+).*$/ ); #Warning: double '_'
        }
        close $BLASTP;
        unlink("$cache/${date}.blastp${r}", "${cache}/${date}_seq2blast.fas${r}") if ( $tmp == 0 || exists($multipleequalblasthits[0]) );
        if ( !exists($multipleequalblasthits[0]) ){
            print "\n$input_order[$r] ...\tNo blast result found above filter thresholds for $right_name\n\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
296
            buildFailureOutputFiles($r, 'No_BLASTp_Result', 'Unavailable', '');
297
298
299
300
301
            undef $original_seq[$r];
            undef $original_names[$r];
            undef $input_order[$r];
            next EACH_SEQ;
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
302
303
304
305
306
    }



    #When multiple equal blast hits  ==> use, and add, every hits
307
308
309
    my ($resultPOS, $resultBOJ) = ('', '');
    my @failureStatus = ('');
    my @intronStatus  = ('');
Sebastien Moretti's avatar
Sebastien Moretti committed
310
    for(my $qq=0; $qq<=$#multipleequalblasthits; $qq++){
311
312
313
314
315
        print "\n${right_name}  -->  [${multipleequalblasthits[$qq]}]\n";


        my @nt_GIs;
        my $intronStep = 0;
316
317
318
        if ( exists($template_NT->{$templ_name}) && $template_NT->{$templ_name} ne '' && $template_NT->{$templ_name} !~ /^My_Seq$/i ){
            downloadSeq($cache, $date, '', '', $template_NT->{$templ_name});
            @nt_GIs = $template_NT->{$templ_name};
319
        }
320
        elsif ( exists($template_NT->{$templ_name}) && $template_NT->{$templ_name} ne '' && $template_NT->{$templ_name} =~ /^My_Seq$/i ){
321
            open(my $MYSEQ, '>', "$cache/My_Seq-$$.fas");
322
            print {$MYSEQ} ">My_Seq-$$ for $templ_name\n".$template_SEQ->{$templ_name}."\n";
323
324
325
326
327
            close $MYSEQ;
            @nt_GIs = "My_Seq-$$";
        }
        else{
        #Prot ACC -> PUID
Sebastien Moretti's avatar
Sebastien Moretti committed
328
            my $protGI = blastPAcc2PGI($multipleequalblasthits[$qq]);
329
330
331
332
333
            if ( $protGI eq '' ){
                print "\tNo protein (prot GI) link found for ${multipleequalblasthits[$qq]} in $right_name\n\n";
                @failureStatus = ('PUI_unavailable', ${multipleequalblasthits[$qq]});
                next;
            }
Sebastien Moretti's avatar
Sebastien Moretti committed
334
335
#           print "\t$protGI\n";

336
        #PUID -> nt GIs & GeneID
Sebastien Moretti's avatar
Sebastien Moretti committed
337
            my ($ntGIs, $geneID) = protGI2NTGIs($protGI, $multipleequalblasthits[$qq]);
338
339
340
341
342
343
344
345
346
347
            if ( $ntGIs eq '' && $geneID eq '' ){
                print "\tNo nucleotide (nt GIs) link found for ${multipleequalblasthits[$qq]} in $right_name\n\n";
                @failureStatus = ('No_nt_link', ${multipleequalblasthits[$qq]});
                next;
            }
            print " linked with [$ntGIs $geneID]\n";


        #Get transcript and contig seq
            @nt_GIs = split(/,/, $ntGIs);
Sebastien Moretti's avatar
Sebastien Moretti committed
348
            downloadSeqFromGIs($cache, $date, '', '', @nt_GIs);
349
350
351
352
        #GeneID -> Chr
            my @geneIDs = split(/,/, $geneID);
            while(<@geneIDs>){
                my $geneID = $_;
Sebastien Moretti's avatar
Sebastien Moretti committed
353
                my ($chr, $amont, $aval) = geneID2Chr($geneID, $multipleequalblasthits[$qq]);
354
355
356
357
                print "\n\tNo gene locus found for ${multipleequalblasthits[$qq]} with $geneID in $right_name\n\n" if ($chr eq '' and $geneID ne '');

                if ( $chr ne '' ){
            #Get Chr seq
Sebastien Moretti's avatar
Sebastien Moretti committed
358
                    downloadSeq($cache, $date, $amont, $aval, $chr);
359
360
361
362
363
364
365
366
367
                    $intronStep = 1;

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


    #Align Nt seq with our protein query seq
368
        my ($POS, $BOJ) = runExonerate($exonerate_exe, $cache, $date, ${input_order[$r]}, ${original_seq[$r]}, $right_name, $multipleequalblasthits[$qq], $r, @nt_GIs);
369

Sebastien Moretti's avatar
Sebastien Moretti committed
370
        my ($POSresult, $bestNucleotide) = prepareResults4CDS($POS, $multipleequalblasthits[$qq], ${original_seq[$r]}, $right_name);
371
372
373
374
375
376
377
378
        $resultPOS .= $POSresult if ( $bestNucleotide ne '' && $POSresult ne '' && $resultPOS !~ /_G_$bestNucleotide/ );

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

        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]_/ || $bestNucleotide =~ /^AC_/) );
Sebastien Moretti's avatar
Sebastien Moretti committed
379
        my ($BOJresult, $bestGenomic) = prepareResults4BOJ($BOJ, $multipleequalblasthits[$qq], ${original_seq[$r]}, $right_name, $intronLess, $nameLess);
380
        $resultBOJ .= $BOJresult      if ( $bestGenomic ne '' && $BOJresult ne '' && $resultBOJ !~ /_G_$bestGenomic[ \-]/ );
381
#    print " $intronStep $POS->{0} [@intronStatus] {$BOJresult} $bestNucleotide - $bestGenomic \t $BOJ->{0}\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
382
383
384
    }


385
    if ( $resultPOS eq '' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
386
387
        buildFailureOutputFiles($r, $failureStatus[1], $failureStatus[0], '')     if ( $failureStatus[0] ne '' );
        buildFailureOutputFiles($r, $multipleequalblasthits[0], 'Failed_Aln', '') if ( $failureStatus[0] eq '' );
388
        next;
Sebastien Moretti's avatar
Sebastien Moretti committed
389
    }
390
391
    elsif ( $resultBOJ eq '' ){
        if ( $failureStatus[0] eq 'PUI_unavailable' || $failureStatus[0] eq 'No_nt_link' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
392
            buildFailureBOJOutputFile($failureStatus[1], $failureStatus[0], $right_name, ${original_seq[$r]});
393
394
        }
        elsif ( $failureStatus[0] eq '' && $intronStatus[0] ne '' ){
Sebastien Moretti's avatar
Sebastien Moretti committed
395
            buildIntronlessBOJOutputFile($intronStatus[1], $intronStatus[0], $intronStatus[2], $right_name, ${original_seq[$r]});
396
397
        }
        else{
Sebastien Moretti's avatar
Sebastien Moretti committed
398
            buildFailureBOJOutputFile($multipleequalblasthits[0], 'Unavailable', $right_name, ${original_seq[$r]});
399
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
400
        createCDSOutputFile($resultPOS, ${original_seq[$r]}, $right_name);
Sebastien Moretti's avatar
Sebastien Moretti committed
401
402
    }
    else {
Sebastien Moretti's avatar
Sebastien Moretti committed
403
404
        createCDSOutputFile($resultPOS, ${original_seq[$r]}, $right_name);
        createBOJOutputFile($resultBOJ, ${original_seq[$r]});
Sebastien Moretti's avatar
Sebastien Moretti committed
405
406
407
408
409
410
411
412
413
    }


    undef $original_seq[$r];
    undef $original_names[$r];
    undef $input_order[$r];
}


Sebastien Moretti's avatar
Sebastien Moretti committed
414
415
checkAndCleanStderrFiles("$cache/${date}_ExonerateError") if ( -e "$cache/${date}_ExonerateError" );
checkOutputFiles();
Sebastien Moretti's avatar
Sebastien Moretti committed
416
unlink("$msa") if ( $converted==1 );
417
my @tmpFiles = glob($cache."/${date}*");
Sebastien Moretti's avatar
Sebastien Moretti committed
418
419
if ( exists($tmpFiles[0]) ){
    mkdir("$cache/${date}_TMP");
420
    move("$_", "$cache/${date}_TMP") while(<@tmpFiles>);
Sebastien Moretti's avatar
Sebastien Moretti committed
421
422
423
424
425
426
427
}


#Ouput Checker


#STDOUT Log output for our server
428
if ( -s "${originalMSA}.cds" || -s "${originalMSA}.out"){
Sebastien Moretti's avatar
Sebastien Moretti committed
429
430
    print "\n\nTERMINATION STATUS: SUCCESS\n";
    print "OUTPUT RESULTS\n";
Sebastien Moretti's avatar
Sebastien Moretti committed
431
432
433
434
435
    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= ${originalMSA}.cdsP.html\n" 
Sebastien Moretti's avatar
Sebastien Moretti committed
436
        if ( Views::Html("${originalMSA}.cdsP") && -s "${originalMSA}.cdsP.html");
Sebastien Moretti's avatar
Sebastien Moretti committed
437
438
439
440
    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
441
442
}
else{
Sebastien Moretti's avatar
Sebastien Moretti committed
443
    failure();
444
    exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
445
446
}

447
exit(0);
Sebastien Moretti's avatar
Sebastien Moretti committed
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
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


=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

492
=item empty: empty completely the cache directory
Sebastien Moretti's avatar
Sebastien Moretti committed
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529

=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

530
=item version 4.0.0
Sebastien Moretti's avatar
Sebastien Moretti committed
531

532
=item on Apr 02nd, 2011
Sebastien Moretti's avatar
Sebastien Moretti committed
533
534
535
536
537
538
539
540
541

=back

=head1 AUTHORS

=over 8

=item Sebastien MORETTI

Sebastien Moretti's avatar
Sebastien Moretti committed
542
=item moretti.sebastien [AT] gmail.com
Sebastien Moretti's avatar
Sebastien Moretti committed
543

544
=item Vital-IT computing center
Sebastien Moretti's avatar
Sebastien Moretti committed
545

546
=item Swiss Institute of Bioinformatics
Sebastien Moretti's avatar
Sebastien Moretti committed
547

548
=item Lausanne, Switzerland
Sebastien Moretti's avatar
Sebastien Moretti committed
549

550
=item http://www.vital-it.ch/
Sebastien Moretti's avatar
Sebastien Moretti committed
551
552
553
554
555
556
557
558

=back

=cut


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

559
# Failure declaration
Sebastien Moretti's avatar
Sebastien Moretti committed
560
sub failure {
561
562
563
    print {*STDERR} "\n\nFATAL\n\nTERMINATION STATUS: FAILURE\n\n";
    return;
}
Sebastien Moretti's avatar
Sebastien Moretti committed
564

Sebastien Moretti's avatar
Sebastien Moretti committed
565
# Check external programs presence in the PATH
566
sub checkExternalSoftware {
Sebastien Moretti's avatar
Sebastien Moretti committed
567
568
569
570
571
572
    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
573
    }
Sebastien Moretti's avatar
Sebastien Moretti committed
574
    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
575
576
}

577
578
# Check cache directory accessibility: Test and change rights, if required, of the cache directory
sub checkCacheAccessibility {
579
    my ($cacheDir) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
580

581
    if ( -d "$cacheDir/" && -w "$cacheDir/" && -x "$cacheDir/" && -r "$cacheDir/" ){
582
        # Good
Sebastien Moretti's avatar
Sebastien Moretti committed
583
    }
584
585
586
    elsif ( -d "$cacheDir/" ){
        if ( -o "$cacheDir/" ){
            chmod 0754, "$cacheDir/";
587
        }
588
        else{
589
590
591
            failure();
            print {*STDERR} "\tPermissions do NOT seem to be valid for the '$cacheDir' cache directory\n\n";
            exit(2);
592
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
593
594
    }
    else{
595
        mkdir "$cacheDir/"; # FIXME: mkdir is not recursive !
596
        if ( -o "$cacheDir/" ){
597
            chmod 0754, "$cacheDir/"; #0754 is in octal, so 754 or '0754' are not good !
598
599
        }
        else{
600
601
602
            failure();
            print {*STDERR} "\tPermissions do NOT seem to be valid for the '$cacheDir' cache directory\n\n";
            exit(2);
603
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
604
    }
605

606
    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
607
608
}

609
# Cache Behavior Management
Sebastien Moretti's avatar
Sebastien Moretti committed
610
sub cacheManagement{
611
    my ($cache, $cacheDir, $inScript) = @_;
612

613
    if ( $cache eq 'empty' ){
614
615
        unlink <$cacheDir/*.fas>;
        unlink <$cacheDir/*_ExonerateError>;
616
617
        unlink("$cacheDir/webblast.log", "$cacheDir/web_tempo.result", "$cacheDir/debug.tempo");
        print "\n\tcache directory [$cacheDir] is empty\n\n";
618
        exit 0;
Sebastien Moretti's avatar
Sebastien Moretti committed
619
    }
620
    elsif ( $cache eq 'old' ){
621
622
623
624
625
        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;
626
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
627
    }
628
    elsif ( $cache eq 'update' || $cache eq 'use' ){
629
630
        # DEFAULT: Limit cached files to less than $cacheStorageTime days old ones for 'update'
        $cache = $cacheDir;
Sebastien Moretti's avatar
Sebastien Moretti committed
631
    }
632
    elsif ( $cache eq 'none' ){
633
        # Do not use cache
Sebastien Moretti's avatar
Sebastien Moretti committed
634
    }
635
636
637
638
    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 $cache );
Sebastien Moretti's avatar
Sebastien Moretti committed
639
640
    }
    else {
641
642
643
        failure();
        print {*STDERR} "\n\tWrong cache argument, or your cache folder '$cacheDir' doesn't seem to exist\n\n";
        exit(3);
Sebastien Moretti's avatar
Sebastien Moretti committed
644
645
646
647
648
649
650
651
652
653
    }

    return($cache);
}

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


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

Sebastien Moretti's avatar
Sebastien Moretti committed
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
# 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';
            $converted = 1;
669
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
670
671
    }
    else {
Sebastien Moretti's avatar
Sebastien Moretti committed
672
        failure();
673
674
        print {*STDERR} "\tThe file \'$msa\' doesn't not seem to be reachable\n\n";
        exit(1);
Sebastien Moretti's avatar
Sebastien Moretti committed
675
676
    }

Sebastien Moretti's avatar
Sebastien Moretti committed
677
    return($msa, $converted);
Sebastien Moretti's avatar
Sebastien Moretti committed
678
679
680
681
}

#Prepare Fasta Header of the query name to incorporate our annotations
sub fastaHeaders4ProtoGene{
682
    my ($QueryName) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
683
684

    $QueryName =~ s/  +/ /g;
685
686
687
688
    my ($cqacc, $cqdesc) = ($QueryName, $QueryName);
    $cqacc  =~ s/^ *([^ ]*) .*$/$1/;
    $cqacc  =~ s/[\s\t]*$//g;
    $cqacc  =  $cqacc.'_G_@@';
Sebastien Moretti's avatar
Sebastien Moretti committed
689
690
691
692
    $cqdesc =~ s/^ *[^ ]* *(.*)$/$1/;
    $cqdesc =~ s/^[\| \.]*//;
    $cqdesc =~ s/[\|\. ]*$//;

693
    $QueryName = $cqacc.$cqdesc;
Sebastien Moretti's avatar
Sebastien Moretti committed
694
695
696
697
698
    return($QueryName);
}

#Prepare Fasta Header to put it on the fasta header of the result outputs
sub getFastaHeaderAnnot{
699
    my ($bestTarget) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
700
701

    #Get annotation from nucleotide file header
702
703
704
    open(my $BEST, '<', "$cache/${bestTarget}.fas");
    my $annot=<$BEST> || '';
    close $BEST;
Sebastien Moretti's avatar
Sebastien Moretti committed
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
    chomp($annot);
    $annot =~ s/^ *[^ ]* //;
    $annot =~ s/[\. ]*$//;
    $annot =~ s/^ *//;
    $annot =~ s/  +/ /g;

    return($annot);
}

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


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

#Direct Reverse translation
sub reverse_trad{
721
    my ($aa) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
722
723
724
725
726
727
728

    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

729
730
731
732
733
734
    my $triplet = 'nnn';
    while( my ($x,$y) =each(%reverse_code) ){
        if ( uc($aa) eq $x ){
            $triplet = $y;
            last;
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
735
736
737
738
739
740
741
742
743
744
    }

    return($triplet);
}

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


####################### webblast #######################
sub launchWebblast{
745
746
    my ($infile, $quiet, $whatBlast, $db, $species, $blast_exe, $outfile) = @_;
    my $blast_at = '';
Sebastien Moretti's avatar
Sebastien Moretti committed
747

748
749
750
    $blast_at    = "-database=$db"                         if ( $whatBlast==0 );
    $blast_at    = "-database=$db -blast_dir=${blast_exe}" if ( $whatBlast==1 );
    $blast_at    = "-database=$db -gigablast=yes"          if ( $whatBlast==2 );
Sebastien Moretti's avatar
Sebastien Moretti committed
751
752
753
754
755
756
757
758
759
760
761
762
763

#    print "             Evalue treshold : 0.05
#             Matrix : BLOSUM62
#             Filter : F
#             **********************
#             For RefSeq Protein db:
#             Blast_identity_threshold : 100
#             Cover threshold : 100
#             **********************
#             For NR Protein db:
#             Blast_identity_threshold : 95
#             Cover threshold : 95";

764
    system("$webblast_exe -program=blastp ${blast_at} -infile=$infile -matrix=BLOSUM62 -evalue=0.05 -method=geneid -filter=Off -organism=$species -identity=100 -cover=100 -hits=10 $quiet -outfile=$outfile");
Sebastien Moretti's avatar
Sebastien Moretti committed
765
766


767
768
769
770
771
772
773
    if ( -z "$outfile" ){ #Another blastP query if no hit with refseq and db was not nr !
        $quiet = '-quiet=on';
        print {*STDERR} "run BLAST.. against NR because there was no acceptable hit against $db\n"
            if ( $whatBlast != 1 && $db ne 'nr' );
        print {*STDERR} "run BLAST.. again, with decreased thresholds, because there was no acceptable hit against $db\n"
            if ( $whatBlast == 1 || $db eq 'nr' );
        $blast_at =~ s/-database=[^\s]*/-database=nr/ if ( $whatBlast != 1 );
774
        system("$webblast_exe -program=blastp ${blast_at} -infile=$infile -matrix=BLOSUM62 -evalue=0.05 -method=geneid -filter=Off -organism=$species -identity=95 -cover=95 -hits=10 $quiet -outfile=$outfile");
Sebastien Moretti's avatar
Sebastien Moretti committed
775
776
    }

777
778
779
780
781
782
783
    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
784

785
    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
786
787
788
789
790
791
792
793
794
795
}


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


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

#Prot ACC -> PUID
sub blastPAcc2PGI{
796
    my ($blastHit) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
797

798
799
800
    my $protGI = '';
    my $count  = 0;
    GET_PUI:
Sebastien Moretti's avatar
Sebastien Moretti committed
801
    for(my $rep=0;$rep <= 4; $rep++){
802
        $count++;
803
804
#        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'");
805
806
807
        open(my $GIP, '<', "$cache/${date}_${blastHit}protGi.tmp");
        PROT_GI:
        while(<$GIP>){
808
            if ( $_ =~ /^.*<Id>(\d+)<\/Id>.*$/ ){
809
810
811
812
813
814
815
                $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
816
    }
817
    unlink("$cache/${date}_${blastHit}protGi.tmp") if ( $tmp == 0 || $protGI ne '' );
Sebastien Moretti's avatar
Sebastien Moretti committed
818
819
820
821
822
823

    return($protGI);
}

#PUID -> nt GIs
sub protGI2NTGIs{
824
    my ($protGI, $blastHit) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
825

826
827
828
829
    my $ntGIs  = '';
    my $geneID = '';
    my $count  = 0;
    GET_NTUI:
Sebastien Moretti's avatar
Sebastien Moretti committed
830
    for(my $rep=0;$rep <= 4; $rep++){
831
        $count++;
Sebastien Moretti's avatar
Sebastien Moretti committed
832
        #nt GIs
833
        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'");
834
835
836
837
838
839
840
841
842
843
844
        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 && $_ =~ /^.*\<Id\>(\d+)\<\/Id\>.*$/ ){
845
846
847
                my $match = $1;
                $ntGIs  .= ",$match" if ( $ntGIs  ne '' && $ntGIs  !~ /,$match,/ && $ntGIs  !~ /^$match,/ );
                $ntGIs   = $match    if ( $ntGIs  eq '' );
848
849
            }
            elsif ( $flag==2 && $_ =~ /^.*\<Id\>(\d+)\<\/Id\>.*$/ ){
850
851
852
                my $match = $1;
                $geneID .= ",$match" if ( $geneID ne '' && $geneID !~ /,$match,/ && $geneID !~ /^$match,/ );
                $geneID  = $match    if ( $geneID eq '' );
853
854
855
856
857
            }
        }
        close $GIN;
        $rep = $rep-15 if ( -z "$cache/${date}_${blastHit}nucleo.tmp" && $count==1 );
        last GET_NTUI if ( $ntGIs ne '' || $geneID ne '' );
Sebastien Moretti's avatar
Sebastien Moretti committed
858
    }
859
    unlink("$cache/${date}_${blastHit}nucleo.tmp") if ( $tmp==0 || ($ntGIs ne '' || $geneID ne '') );
Sebastien Moretti's avatar
Sebastien Moretti committed
860

861
    return($ntGIs, $geneID);
Sebastien Moretti's avatar
Sebastien Moretti committed
862
863
864
}

sub geneID2Chr{
865
    my ($geneID, $blastHit) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
866

867
868
869
870
    my $chr   = '';
    my ($amont, $aval) = ('', '');
    my $count = 0;
    GET_CHR:
Sebastien Moretti's avatar
Sebastien Moretti committed
871
    for(my $rep=0;$rep <= 8; $rep++){
872
        $count++;
Sebastien Moretti's avatar
Sebastien Moretti committed
873
        #Gene Acc
874
875
#        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'");
876
877
878
879
880
881
882
        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 );
            }
883
884
885
#            if ( $_   =~ /\<Entrezgene_locus\>/ && $flag==0 ){
            if ( $_   =~ /<Item Name="GenomicInfoType" Type="Structure">/ && $flag==0 ){
               $flag = 1;
886
            }
887
888
889
890
891
892
893
#            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
894
            elsif ( $_ =~ /<Item Name="ChrAccVer" Type="String">([^<]+?)\.?\d*<\/Item>/ && $flag==1 ){
895
896
897
                $chr   = $1;
                $flag  = 3;
            }
898
899
#            elsif ( $_ =~ /\<Seq-interval_from\>(\d+)\<\/Seq-interval_from\>/ && $flag==3 ){
            elsif ( $_ =~ /<Item Name="ChrStart" Type="Integer">(\d+)<\/Item>/ && $flag==3 ){
900
901
902
                $amont = $1;
                $flag  = 4;
            }
903
904
#            elsif ( $_ =~ /\<Seq-interval_to\>(\d+)\<\/Seq-interval_to\>/ && $flag==4 ){
            elsif ( $_ =~ /<Item Name="ChrStop" Type="Integer">(\d+)<\/Item>/ && $flag==4 ){
905
906
907
908
909
910
911
                $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
912
    }
913
914
915
916
    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
917
    }
918
919
920
921
922
923
924
925
926
    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
927
    }
928
929
    $amont = 1 if ( $amont ne '' && $amont <= 0 );
    $aval  = 1 if ( $aval ne ''  && $aval <= 0 );
Sebastien Moretti's avatar
Sebastien Moretti committed
930

931
    return($chr, $amont, $aval);
Sebastien Moretti's avatar
Sebastien Moretti committed
932
933
934
}

sub downloadSeqFromGIs{
935
    my ($cache, $date, $amont, $aval, @acc) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
936

937
    GET_SEQ:
Sebastien Moretti's avatar
Sebastien Moretti committed
938
    for(my $a=0; $a<=$#acc; $a++){
939
940
941
942
        my $whatNumber = 0;
        if ( $amont =~ /^\d+$/ && $aval =~ /^\d+$/ ){
            GET_CHR_SEQ:
            for(my $rep=0;$rep <= 4; $rep++){
943
                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 ) );
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
                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++){
967
                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 ) );
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
                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:/ || $_ =~ /[<>]/) );
                }
                close $CIBLE;
                $whatNumber++;
                last GET_OTHER_SEQ if ( $whatNumber==20 );
                if ( $counter != 1 ){
                    $rep=$rep-1;
                    unlink("$cache/${acc[$a]}.fas");
                }
            }
        }
Sebastien Moretti's avatar
Sebastien Moretti committed
987
    }
988
989

    return;
Sebastien Moretti's avatar
Sebastien Moretti committed
990
991
992
}

sub downloadSeq{
993
    my ($cache, $date, $amont, $aval, @acc) = @_;
Sebastien Moretti's avatar
Sebastien Moretti committed
994

995
996
997
998
    my $cp   = 0;
    my $from = $amont;
    my $to   = $aval;
    DOWNL_SEQ:
Sebastien Moretti's avatar
Sebastien Moretti committed
999
    for(my $a=0; $a<=$#acc; $a++){
1000
        my $pacc2puid = $acc[$a];