0
|
1 #!perl
|
|
2
|
|
3 ## script : bank_inhouse.pl
|
|
4 #=============================================================================
|
|
5 # Included modules and versions
|
|
6 #=============================================================================
|
|
7 ## Perl modules
|
|
8 use strict ;
|
|
9 use warnings ;
|
|
10 use Carp qw (cluck croak carp) ;
|
|
11
|
|
12 use Data::Dumper ;
|
|
13 use Getopt::Long ;
|
|
14 use POSIX ;
|
|
15 use List::Util qw( min max );
|
|
16 use FindBin ; ## Allows you to locate the directory of original perl script
|
|
17
|
|
18 ## Specific Modules
|
|
19 use lib $FindBin::Bin ;
|
|
20 my $binPath = $FindBin::Bin ;
|
|
21 use lib::bih qw( :ALL ) ;
|
|
22
|
|
23 ## PFEM Perl Modules
|
|
24 use lib::conf qw( :ALL ) ;
|
|
25 use lib::csv qw( :ALL ) ;
|
|
26
|
|
27 ## Initialized values
|
|
28 my ( $help ) = ( undef ) ;
|
|
29 my ( $mass ) = ( undef ) ;
|
|
30 my ( $masses_file, $nbline_header, $col_mass ) = ( undef, undef, undef ) ;
|
|
31 my ( $col_rt, $manual_rt, $rt_delta, $mz_delta_type, $mz_delta, $rtdb ) = ( undef, undef, undef, undef, undef, undef ) ;
|
|
32 my ( $mode, $tissues, $bank_in, $bank_name, $col_mzdb) = ( undef, undef, undef, undef, undef ) ;
|
|
33 my ( $out_tab, $out_html, $out_xls, $out_json, $out_full ) = ( undef, undef, undef, undef, undef ) ;
|
|
34 my ($rest_mode) = ('no') ;
|
|
35
|
|
36 my ( $verbose ) = ( 2 ) ; ## verbose level is 3 for debugg
|
|
37
|
|
38 #=============================================================================
|
|
39 # Manage EXCEPTIONS
|
|
40 #=============================================================================
|
|
41
|
|
42 &GetOptions ( "h" => \$help, # HELP
|
|
43 "masse:s" => \$mass, ## option : one masse
|
|
44 "input:s" => \$masses_file, ## option : path to the input
|
|
45 "rest:s" => \$rest_mode, ## option : allow uses of peakforest rest service
|
|
46 "nbheader:i" => \$nbline_header, ## numbre of header line present in file
|
|
47 "colrt:i" => \$col_rt, ## Column retention time for retrieve formula/masses list in tabular file
|
|
48 "rt:s" => \$manual_rt, ## Retention time for manual masses list
|
|
49 "rt_delta:f" => \$rt_delta, ## Delta for retention time
|
|
50 "colmass:i" => \$col_mass, ## Column id for retrieve formula list in tabular file
|
|
51 "mz_delta:s" => \$mz_delta_type, ## Delta type for masses
|
|
52 "mass_delta:f" => \$mz_delta, ## Delta for masses
|
|
53 "mode:s" => \$mode, ## Molecular species (positive/negative/neutral)
|
|
54 "tissues:s" => \$tissues, ## Restricted to certain characterization and localization (blood/plasma/urine/peptide)
|
|
55 "bank_in:s" => \$bank_in, ## option : path to a personal data bank
|
|
56 "bank_name:s" => \$bank_name, ## option : name of the bank
|
|
57 "mzdb:i" => \$col_mzdb, ## Column number in which is the masses of the personal data bank
|
|
58 "rtdb:i" => \$rtdb, ## Column number in which is the retention time of the personal data bank
|
|
59 "outputJson:s" => \$out_json, ## option : path to the ouput (results)
|
|
60 "outputXls:s" => \$out_xls, ## option : path to the ouput (tabular : results )
|
|
61 "outputTab:s" => \$out_tab, ## path to the ouput (tabular : input+results (1 col) )
|
|
62 "outputView:s" => \$out_html, ## option : path to the results view (output2)
|
|
63 "outputFull:s" => \$out_full, ## path to the ouput (tabular : input+results (x col) )
|
|
64 ) ;
|
|
65
|
|
66 #=============================================================================
|
|
67 # EXCEPTIONS
|
|
68 #=============================================================================
|
|
69 ## if you put the option -help or -h function help is started
|
|
70 if ( defined $help ) { &help() ; }
|
|
71
|
|
72 #=============================================================================
|
|
73 # MAIN SCRIPT
|
|
74 #=============================================================================
|
|
75
|
|
76
|
|
77 print "The bank_inhouse program is launched as:\n";
|
|
78 print "./bank_inhouse.pl " ;
|
|
79 print "--h " if (defined $help) ;
|
|
80 print "--masse $mass " if (defined $mass) ;
|
|
81 print "--input $masses_file " if (defined $masses_file) ;
|
|
82 print "--rest $rest_mode " if (defined $rest_mode) ;
|
|
83 print "--nbheader $nbline_header " if (defined $nbline_header) ;
|
|
84 print "--colrt $col_rt " if (defined $col_rt) ;
|
|
85 print "--rt $manual_rt " if (defined $manual_rt) ;
|
1
|
86 print "--rt_delta $rt_delta " if (defined $rt_delta) ;
|
0
|
87 print "--colmass $col_mass " if (defined $col_mass) ;
|
|
88 print "--mz_delta $mz_delta_type " if (defined $mz_delta_type) ;
|
|
89 print "--mass_delta $mz_delta " if (defined $mz_delta) ;
|
|
90 print "--mode $mode " if (defined $mode) ;
|
|
91 print "--tissues $tissues " if (defined $tissues) ;
|
|
92 print "--bank_in $bank_in " if (defined $bank_in) ;
|
|
93 print "--bank_name $bank_name " if (defined $bank_name) ;
|
|
94 print "--mzdb $col_mzdb " if (defined $col_mzdb) ;
|
|
95 print "--rtdb $rtdb " if (defined $rtdb) ;
|
|
96 print "--outputJson $out_json " if (defined $out_json) ;
|
|
97 print "--outputXls $out_xls " if (defined $out_xls) ;
|
|
98 print "--outputTab $out_tab " if (defined $out_tab) ;
|
|
99 print "--outputView $out_html " if (defined $out_html) ;
|
|
100 print "--outputFull $out_full " if (defined $out_full) ;
|
|
101 print "\n" ;
|
|
102
|
|
103
|
|
104
|
|
105 ## -------------- Conf file ------------------------ :
|
|
106 my ( $CONF, $PF_CONF ) = ( undef, undef ) ;
|
|
107 foreach my $conf ( <$binPath/*.cfg> ) {
|
|
108 my $oConf = lib::conf::new() ;
|
|
109 $CONF = $oConf->as_conf($conf) ;
|
|
110 }
|
|
111
|
|
112 ## PForest dedicated conf
|
|
113 if ((defined $rest_mode ) and ( $rest_mode eq "yes" )) {
|
|
114 foreach my $ini ( <$binPath/*.ini> ) {
|
|
115 my $oConf = lib::conf::new() ;
|
|
116 $PF_CONF = $oConf->as_conf($ini) ;
|
|
117 }
|
|
118 }
|
|
119
|
|
120 ## --------------- Global parameters ---------------- :
|
|
121 my ( $masses, $rt, $results, $header_choice ) = ( undef, undef, undef, undef, undef ) ; ##############
|
|
122 my ( $complete_rows, $nb_pages_for_html_out ) = ( undef, 1 ) ;
|
|
123
|
|
124 my $oBih = lib::bih::new() ;
|
|
125
|
|
126 ## --------------- retrieve input data -------------- :
|
|
127
|
|
128 ## manage manual mass(es)
|
|
129 if ( ( defined $mass ) and ( $mass ne "" ) ) {
|
|
130 $mass =~ tr/,/./ ;
|
|
131 @$masses = split(' ', $mass);
|
|
132 if ( ( defined $manual_rt ) and ( $manual_rt ne "" ) ) {
|
|
133 $manual_rt =~ tr/,/./ ;
|
|
134 @$rt = split(' ', $manual_rt) ;
|
|
135 if ( $#$rt != $#$masses) { croak "You have not given the same number of masses and retention time\n" ; }
|
|
136 }
|
|
137 } ## END IF
|
|
138
|
|
139 ## manage csv file containing list of masses
|
|
140 elsif ( ( defined $masses_file ) and ( $masses_file ne "" ) and ( -e $masses_file ) ) {
|
|
141 ## parse all csv for later : output csv build
|
|
142 my $ocsv_input = lib::csv->new() ;
|
|
143 my $complete_csv = $ocsv_input->get_csv_object( "\t" ) ;
|
|
144 $complete_rows = $ocsv_input->parse_csv_object($complete_csv, \$masses_file) ;
|
|
145
|
|
146 ## parse csv ids and masses
|
|
147 my $ocsv = lib::csv->new() ;
|
|
148 my $csv = $ocsv->get_csv_object( "\t" ) ;
|
1
|
149
|
|
150 if ( ( defined $nbline_header ) and ( $nbline_header > 0 ) ) { $header_choice = 'yes' ; }
|
|
151 else{ $header_choice = 'no' ; }
|
|
152
|
|
153 ## retrieve mz values on csv
|
|
154 $masses = $ocsv->get_value_from_csv_multi_header( $csv, $masses_file, $col_mass, $header_choice, $nbline_header ) ;
|
|
155
|
|
156 ## retrieve rt values on csv
|
0
|
157 if ( ( defined $col_rt ) and ( $col_rt ne "" ) ) {
|
|
158 $rt = $ocsv->get_value_from_csv_multi_header( $csv, $masses_file, $col_rt, $header_choice, $nbline_header ) ; ## retrieve rt values on csv
|
|
159 }
|
|
160 }
|
|
161
|
|
162 # --------------- retrieve personal data bank input -------------- :
|
|
163
|
|
164 my ($complete_bank, $bank_head) = (undef, undef) ;
|
|
165
|
|
166 ## check up $bank_in and parse
|
|
167 if ( ( defined $bank_in ) and ( $bank_in ne '' ) and ( -e $bank_in ) and ( defined $col_mzdb ) and ( $col_mzdb ne '' ) ) {
|
|
168 ## parse csv bank_in
|
|
169 my $ocsv_input = lib::csv->new() ;
|
|
170 my $complete_csv = $ocsv_input->get_csv_object( "\t" ) ;
|
|
171 ($complete_bank, $bank_head)= $oBih->parse_bank_interest($complete_csv, \$bank_in, $col_mzdb-1) ;
|
|
172 }
|
|
173 ## manage rest service of PeakForest and set $masses and $rt with given databases
|
|
174 elsif ( ( defined $rest_mode ) and ( $rest_mode eq "yes" ) ) {
|
|
175 $bank_name = "PeakForest";
|
|
176 if ( ( $PF_CONF->{PF_GLOBAL_TOKEN} ne "" ) and ( $PF_CONF->{PF_WS_URL} ne "" ) ) {
|
|
177 ## init
|
|
178 my $opfws = lib::bih::new() ;
|
|
179 ## get min and max values of user query
|
|
180 my ($min, $max) = (0, 0) ;
|
|
181
|
|
182 if ( ( defined $masses ) and ( scalar (@{$masses}) > 0 ) ) {
|
|
183 $min = min @{$masses} ;
|
|
184 $max = max @{$masses} ;
|
1
|
185
|
0
|
186 my ( $min_delta, undef ) = $opfws->mz_delta_conversion(\$min, \$mz_delta_type, \$mz_delta) ;
|
|
187 my ( undef, $max_delta ) = $opfws->mz_delta_conversion(\$max, \$mz_delta_type, \$mz_delta) ;
|
|
188
|
|
189 ## get clean range database json from PForest WS :
|
|
190 my $pf_json = $opfws->db_pforest_get_clean_range($PF_CONF->{PF_WS_URL}, $PF_CONF->{PF_REST_QUERY_CLEAN_RANGE}, $$min_delta, $$max_delta, $mode) ;
|
|
191 if (defined $pf_json) {
|
|
192 ($complete_bank, $bank_head) = $opfws->map_pfjson_bankobject($pf_json) ;
|
|
193 }
|
|
194 } ## End IF
|
|
195 else {
|
|
196 croak "Fatal error : your mass list is empty\n" ;
|
|
197 }
|
|
198 } ## End IF PForest param
|
|
199 else { croak "Can't work : missing a PForest REST paramater\n" ; }
|
|
200 } ## End ELSIF PForest
|
|
201 elsif ( ( defined $CONF->{'INHOUSE_BANK'} ) and ( $CONF->{'INHOUSE_BANK'} ne '' ) ) {
|
1
|
202
|
|
203 $col_mzdb = $CONF->{'INHOUSE_BANK_MZ_COLUMN'} ;
|
|
204 $bank_name = $CONF->{'INHOUSE_BANK_NAME'} ;
|
|
205 my $html_file = $binPath.'/'.$CONF->{'INHOUSE_BANK'} ;
|
|
206
|
0
|
207 if ( (defined $col_rt) or (defined $manual_rt) or (defined $rt_delta) ) { croak "No retention time in the internal bank, please use your own bank\n" ; }
|
1
|
208
|
0
|
209 if ( -e $html_file ) {
|
|
210 ## parse all csv for later : output csv build
|
|
211 my $ocsv_input = lib::csv->new() ;
|
|
212 my $complete_csv = $ocsv_input->get_csv_object( "\t" ) ;
|
|
213 ($complete_bank, $bank_head)= $oBih->parse_bank_interest($complete_csv, \$html_file, $col_mzdb-1) ;
|
|
214 }
|
|
215 else { croak "Can't work : no local inhouse bank name '$html_file'\n" ; }
|
|
216 }
|
|
217 else { croak "Can't work : missing an inhouse bank\n" ; }
|
|
218
|
|
219 ## ---------------- launch queries -------------------- :
|
|
220
|
|
221 my $ionization = undef ;
|
|
222 my $characterization = undef ;
|
|
223 my $valH = undef ;
|
|
224 my $bank = undef ;
|
|
225 my $search_condition = undef ;
|
|
226
|
|
227 #($valH, $ionization)= $oBih->manage_mode($mode, $CONF->{'PROTON_MZ'}, $CONF->{'ELECTRON_MZ'}) ;
|
|
228
|
|
229 ######### Manage Tissues ##############################################################################
|
|
230 #if ( ( !defined $tissues ) or ( $tissues eq '' ) or ( $tissues eq 'none' ) ) {
|
|
231 # $characterization = ['None'];
|
|
232 #} else { @{$characterization} = split(",", $tissues); }
|
|
233 ######### End #########################################################################################
|
|
234
|
|
235 #if ( (@$ionization = (['None'])) and (@$characterization = (['None']))) { $bank = $complete_bank; }
|
|
236 #else { $bank = FILTER LA BANK; }
|
|
237 $bank = $complete_bank;
|
|
238 $valH = 0 ;
|
|
239
|
|
240 #if ( ( defined $mz_delta ) and ( $mz_delta > 0 ) and ( defined $ionization ) and ( defined $characterization ) and ( defined $masses ) and ( defined $ids ) and ( defined $bank ) ) {
|
|
241 # $search_condition = "Search params : Molecular specie = $mode / delta = $mz_delta / characterization and localization = $tissues \n bank file = $bank_in" ;
|
|
242
|
|
243 if ( ( defined $mz_delta ) and ( $mz_delta >= 0 ) and ( defined $masses ) and ( defined $bank ) ) {
|
|
244 ## prepare masses list and execute query
|
|
245 $results = [] ;
|
|
246 if ($mz_delta != 0){
|
|
247 my @sort_masses_bank = sort { $a <=> $b } keys(%$bank);
|
|
248
|
|
249 my $compt_masses = 0;
|
|
250 foreach my $mz (@$masses) {
|
|
251 $compt_masses++;
|
1
|
252 my ($MZmessage) = $oBih->check_interval($mz, $CONF->{'BANK_MZ_MIN'}, $CONF->{'BANK_MZ_MAX'}) ;
|
0
|
253 if ( $MZmessage eq 'OK' ){
|
|
254 my ( $min, $max ) = $oBih->mz_delta_conversion(\$mz, \$mz_delta_type, \$mz_delta) ;
|
|
255
|
|
256 my ($marj_inf) = $oBih->dichotomi_search(\@sort_masses_bank, $min) ;
|
|
257 my ($marj_sup) = $oBih->dichotomi_search(\@sort_masses_bank, $max) ;
|
|
258
|
|
259 my $result = [];
|
|
260 if($$marj_inf != $$marj_sup){
|
|
261 if ($$marj_inf == -1){ $$marj_inf=0; }
|
|
262 for (my $i=$$marj_inf; $i<$$marj_sup; $i++){
|
|
263 my $bank_tmp = $$bank{$sort_masses_bank[$i]};
|
|
264 if ( ( defined $rt ) and ( $rt ne "" ) ) {
|
1
|
265 my ($RTmessage) = $oBih->check_interval($$rt[$compt_masses-1], $CONF->{'BANK_RT_MIN'}, $CONF->{'BANK_RT_MAX'}) ;
|
0
|
266 if ( $RTmessage eq 'OK' ){
|
|
267 my $bank_rt = [];
|
|
268 for(my $nb_rt=0; $nb_rt<=$#$bank_tmp; $nb_rt++){
|
1
|
269 my ($RTbank) = $oBih->check_interval($bank_tmp->[$nb_rt]->[$rtdb-1], $CONF->{'BANK_RT_MIN'}, $CONF->{'BANK_RT_MAX'}) ;
|
0
|
270 my ($RTsearch) = $oBih->check_interval($bank_tmp->[$nb_rt]->[$rtdb-1], $$rt[$compt_masses-1]-$rt_delta, $$rt[$compt_masses-1]+$rt_delta) ;
|
|
271 if ( ($RTbank eq 'OK') and ($RTsearch eq 'OK') ){
|
|
272 push (@$result, $bank_tmp->[$nb_rt]) ;
|
|
273 }
|
|
274 elsif ($RTbank ne 'OK'){ croak "At least one retention time in bank is not valid : $RTbank\n" ; }
|
|
275 }
|
|
276 }
|
|
277 else { croak "The $compt_masses th analyzed retention time : $RTmessage\n" ; }
|
|
278 }
|
|
279 else { push (@$result, @$bank_tmp) ; }
|
|
280 }
|
|
281 }
|
|
282 else{ $result = []; }
|
|
283 push (@$results, [@$result]) ;
|
|
284 }
|
|
285 else { croak "The $compt_masses th value : $MZmessage\n" ; }
|
|
286 print Dumper $results if ($verbose>2) ;
|
|
287 } ## End FOREACH MZ queried
|
|
288 } ## End IF
|
|
289 else{
|
|
290 my $result = [];
|
|
291 my $compt_masses = 0;
|
|
292 foreach my $mz (@$masses) {
|
|
293 $compt_masses++;
|
1
|
294 my ($MZmessage) = $oBih->check_interval($mz, $CONF->{'BANK_MZ_MIN'}, $CONF->{'BANK_MZ_MAX'}) ;
|
0
|
295 if ( $MZmessage eq 'OK' ){
|
|
296 if ($$bank{$mz}){
|
|
297 my $bank_tmp = $$bank{$mz};
|
|
298 if ( ( defined $rt ) and ( $rt ne "" ) ) {
|
1
|
299 my ($RTmessage) = $oBih->check_interval($$rt[$compt_masses-1], $CONF->{'BANK_RT_MIN'}, $CONF->{'BANK_RT_MAX'}) ;
|
0
|
300 if ( $RTmessage eq 'OK' ){my $bank_rt = [];
|
|
301 for(my $nb_rt=0; $nb_rt<=$#$bank_tmp; $nb_rt++){
|
1
|
302 my ($RTbank) = $oBih->check_interval($bank_tmp->[$nb_rt]->[$rtdb-1], $CONF->{'BANK_RT_MIN'}, $CONF->{'BANK_RT_MAX'}) ;
|
0
|
303 my ($RTsearch) = $oBih->check_interval($bank_tmp->[$nb_rt]->[$rtdb-1], $$rt[$compt_masses-1]-$rt_delta, $$rt[$compt_masses-1]+$rt_delta) ;
|
|
304 if ( ($RTbank eq 'OK') and ($RTsearch eq 'OK') ){
|
|
305 push (@$bank_rt, $bank_tmp->[$nb_rt]) ;
|
|
306 }
|
|
307 elsif ($RTbank ne 'OK'){ croak "At least one retention time in bank is not valid : $RTbank\n" ; }
|
|
308 }
|
|
309 if ($#$bank_rt>=0){ ## If there is at least one result
|
|
310 push (@$result, $bank_rt) ;
|
|
311 } ## else $bank_rt = [] as defined
|
|
312 }
|
|
313 else { croak "The $compt_masses th analyzed retention time : $RTmessage\n" ; }
|
|
314 }
|
|
315 else { push (@$result, [$bank_tmp]) ; }
|
|
316 }
|
|
317 else { push (@$result, [] ) ; }
|
|
318 }
|
|
319 else { croak "The $compt_masses th mass : $MZmessage\n" ; }
|
|
320 }
|
|
321 $results = [@$results, @$result] ;
|
|
322 }
|
|
323 }
|
|
324 else {
|
|
325 croak "Can't work : missing paramaters (list of ids, masses, delta, ionization, characterization or inhouse_bank)\n" ;
|
|
326 } ## end ELSE
|
|
327
|
|
328
|
1
|
329 if ( ( defined $rt ) and ( $rt ne "" ) ) { unshift (@$bank_head, ("MzDelta_Query-Bank(".$mz_delta.$mz_delta_type.")", "MzBank", "RtQuery", "RtDelta_Query-Bank(".$rt_delta."min.)", "RtBank")) ; }
|
|
330 else { unshift (@$bank_head, ("MzDelta_Query-Bank(".$mz_delta.$mz_delta_type.")", "MzBank")) ; }
|
0
|
331
|
|
332
|
|
333 ## -------------- Produce JSON output ------------------ :
|
|
334 if ( ( defined $out_json ) and ( defined $results ) ) {
|
|
335 open(JSON, '>:utf8', "$out_json") or die "Cant' create the json file\n" ;
|
|
336 print JSON Dumper $results;
|
|
337 close(JSON) ;
|
|
338 } ## END IF
|
|
339 else {
|
|
340 #croak "Can't create a json output for Massbank : no result found or your output file is not defined\n" ;
|
|
341 }
|
|
342
|
|
343 ## -------------- Produce HTML output ------------------ :
|
|
344
|
|
345 #if ( ( defined $out_html ) and ( defined $results ) ) {
|
|
346 # ## Uses N mz and theirs entries per page (see config file).
|
|
347 # # how many pages you need with your input mz list?
|
|
348 # $nb_pages_for_html_out = ceil( scalar(@{$masses} ) / $CONF->{HTML_ENTRIES_PER_PAGE} ) ;
|
|
349 #
|
|
350 # my $oHtml = lib::bih::new() ;
|
|
351 # my ($tbody_object) = $oHtml->set_html_tbody_object( $nb_pages_for_html_out, $CONF->{HTML_ENTRIES_PER_PAGE} ) ;
|
|
352 # ($tbody_object) = $oHtml->add_mz_to_tbody_object($tbody_object, $CONF->{HTML_ENTRIES_PER_PAGE}, $masses, $ids) ;
|
|
353 # ($tbody_object) = $oHtml->add_entries_to_tbody_object($tbody_object, $CONF->{HTML_ENTRIES_PER_PAGE}, $masses, $results) ;
|
|
354 # print Dumper $tbody_object;exit;
|
|
355 # my $output_html = $oHtml->write_html_skel(\$out_html, $tbody_object, $nb_pages_for_html_out, $search_condition, $CONF->{'HTML_TEMPLATE'}, $CONF->{'JS_GALAXY_PATH'}, $CONF->{'CSS_GALAXY_PATH'}) ;
|
|
356 #
|
|
357 #} ## END IF
|
|
358 #else { croak "Can't create a HTML output for HMDB : no result found or your output file is not defined\n" ; }
|
|
359
|
|
360 ## -------------- Produce XLS like output ------------------ :
|
|
361 if ( ( defined $out_xls ) and ( defined $results ) ) {
|
|
362 my $oxls = lib::bih::new() ;
|
|
363 $oxls->write_excel_like_mass($masses, $mz_delta_type, $mz_delta, $col_mzdb, $rt, $rt_delta, $rtdb, $results, $out_xls, $bank_head) ;
|
|
364 } ## END IF
|
|
365 else { croak "Can't create a tabular output for BiH (like xls): no result found or your output file is not defined\n" ; }
|
|
366
|
|
367 ## -------------- Produce CSV output ------------------ :
|
|
368 if (defined $results) {
|
|
369 if ( defined $masses_file ) { # produce a csv based on METLIN format
|
|
370 if ( defined $out_tab ) {
|
|
371 my $sep = "#";
|
|
372 my $ocsv = lib::bih::new() ;
|
|
373 my $lm_matrix = undef ;
|
|
374 if ( ( $header_choice eq 'yes' ) and ( defined $nbline_header ) and ( $nbline_header > 0 )) {
|
|
375 my $header = join("$sep", @$bank_head);
|
|
376 my $out_head = "BiH_".$bank_name."_(".$header.")";
|
1
|
377 $lm_matrix = $ocsv->set_bih_matrix_object($out_head, $masses, $col_mzdb, $results, $rt, $rtdb, $bank_head, $sep ) ;
|
|
378 $lm_matrix = $ocsv->add_bih_matrix_to_input_matrix($complete_rows, $lm_matrix, $nbline_header-1) ;
|
0
|
379 }
|
|
380 elsif ( ( $header_choice eq 'no' ) or ( $nbline_header <= 0 )) {
|
1
|
381 $lm_matrix = $ocsv->set_bih_matrix_object(undef, $masses, $col_mzdb, $results, $rt, $rtdb, $bank_head, $sep ) ;
|
|
382 $lm_matrix = $ocsv->add_bih_matrix_to_input_matrix($complete_rows, $lm_matrix, 0) ;
|
0
|
383 }
|
|
384 else { croak "The number of header line is not identifiable\n" ; }
|
|
385 $ocsv->write_csv_skel(\$out_tab, $lm_matrix) ;
|
|
386 }
|
|
387 else { croak "Can't create a tabular output for BiH : your output file is not defined\n" ; }
|
|
388 }
|
|
389 elsif (defined $mass) { } ## no csv output for manual masses
|
|
390 else { croak "the input format is not identifiable or your output file is not defined\n" ; }
|
|
391 } ## END IF
|
|
392 else { croak "Can't create a tabular output for BiH : no result found\n" ; }
|
|
393
|
|
394 ## -------------- Produce Full output ------------------ :
|
|
395 if (defined $results) {
|
|
396 if ( defined $masses_file ) {
|
|
397 if ( defined $out_full ) {
|
|
398 my $sep = "\t";
|
|
399 my $ofull = lib::bih::new() ;
|
|
400 if ( ( $header_choice eq 'yes' ) and ( defined $nbline_header ) and ( $nbline_header > 0 )) {
|
|
401 $ofull->write_full_excel_like($complete_rows, $sep, $masses, $mz_delta_type, $mz_delta, $col_mzdb, $rt, $rt_delta, $rtdb, $results, $out_full, $nbline_header, $bank_head, "BiH_$bank_name") ;
|
|
402 }
|
|
403 elsif ( ( $header_choice eq 'no' ) or ( $nbline_header <= 0 )) {
|
|
404 $ofull->write_full_excel_like($complete_rows, $sep, $masses, $mz_delta_type, $mz_delta, $col_mzdb, $rt, $rt_delta, $rtdb, $results, $out_full, 0, undef, undef) ;
|
|
405 }
|
|
406 else { croak "The number of header line is not identifiable\n" ; }
|
|
407 }
|
|
408 }
|
|
409 elsif (defined $mass) { } ## no csv output for manual masses
|
|
410 else { croak "the input format is not identifiable or your output file is not defined\n" ; }
|
|
411 } ## END IF
|
|
412 else { croak "Can't create a full output for BiH : no result found\n" ; }
|
|
413
|
|
414
|
|
415 #====================================================================================
|
|
416 # Help subroutine called with -h option
|
|
417 # number of arguments : 0
|
|
418 # Argument(s) :
|
|
419 # Return : 1
|
|
420 #====================================================================================
|
|
421 sub help {
|
|
422 print STDERR "
|
|
423 bank_inhouse
|
|
424
|
|
425 # bank_inhouse is a script to query a in house bank (file) using chemical mass and return a list of common names.
|
|
426 # Input : mass or list of masses
|
1
|
427 # Author : Marion LANDI and Franck Giacomoni (as maintainers)
|
|
428 # Email : franck.giacomoni\@inra.fr
|
|
429 # Version : 1.2.1
|
0
|
430 # Created : 15/10/2014
|
1
|
431 # Updated : 24/01/2019
|
0
|
432 USAGE :
|
|
433 bank_inhouse.pl -masse [mass]
|
|
434 -tolerance [Delta of mass (Da)] -mode [Ionization type: positive/negative/neutral]
|
|
435 -tissues [Restricted to certain tissues] -- optionnal
|
|
436 -bank_in [file for in personal house bank] -- optionnal
|
|
437 -outputTab [output file tabular format] -outputView [output file html format]
|
|
438 -verbose
|
|
439 OR
|
|
440 bank_inhouse.pl -input [path to list of masses file]
|
|
441 -nbHeader [nb of lines of headers : 0-n] -colId [Ids colunm number in input file] -colmass [masses col] -colrt [RT col]
|
|
442 -tolerance [Delta of mass (Da)] -mode [molecular type: positive/negative/neutral]
|
|
443 -bank_in [file for in house bank]
|
|
444 -tissues [Restricted to certain tissues]
|
|
445 -outputTab [output file tabular format] -outputView [output file html format]
|
|
446
|
|
447 OR
|
|
448 bank_inhouse.pl -masse [mass]
|
|
449 -rest [get the PeakForest ref db : yes|no]
|
|
450 -mz_delta [Delta type for mass (Da or PPM)] -mass_delta [Delta for mass (Da or PPM)] -mode [molecular type: positive/negative/neutral]
|
|
451 -outputXls [output xls format]
|
|
452
|
|
453 Others output formats:
|
|
454 -outputJson [output json format] -outputFull [output format is input+res]
|
|
455 ";
|
|
456 exit(1);
|
|
457 }
|
|
458
|
|
459 ## END of script - M Landi
|
|
460
|
|
461 __END__
|
|
462
|
|
463 =head1 NAME
|
|
464
|
|
465 bank_inhouse.pl -- script for
|
|
466
|
|
467 =head1 USAGE
|
|
468
|
|
469 bank_inhouse.pl -precursors -arg1 [-arg2]
|
|
470 or bank_inhouse.pl -help
|
|
471
|
|
472 =head1 SYNOPSIS
|
|
473
|
|
474 This script manage ...
|
|
475
|
|
476 =head1 DESCRIPTION
|
|
477
|
|
478 This main program is a ...
|
|
479
|
|
480 =over 4
|
|
481
|
|
482 =item B<function01>
|
|
483
|
|
484 =item B<function02>
|
|
485
|
|
486 =back
|
|
487
|
|
488 =head1 AUTHOR
|
|
489
|
|
490 Marion LANDI E<lt>marion.landi@clermont.inra.frE<gt>
|
1
|
491 Frank Giacomoni E<lt>franck.giacomoni@inra.frE<gt>
|
0
|
492
|
|
493 =head1 LICENSE
|
|
494
|
|
495 This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
|
|
496
|
|
497 =head1 VERSION
|
|
498
|
1
|
499 version 1.1.1 : 15 / 10 / 2014
|
|
500
|
|
501 version 1.1.2 : 21 / 11 / 2014
|
0
|
502
|
1
|
503 version 1.1.3 : 13 / 11 / 2018
|
|
504
|
|
505 version 1.2.0 : 12 / 12 / 2018
|
|
506
|
|
507 version 1.2.1 : 24 / 01 / 2019 - new version with no more rest methods
|
|
508
|
0
|
509
|
|
510 =cut
|