Mercurial > repos > fgiacomoni > hmdb_ms_search
comparison lib/hmdb.pm @ 15:beb9619263b0 draft
Master branch Updating - - Fxx
| author | fgiacomoni |
|---|---|
| date | Wed, 16 Jan 2019 10:18:18 -0500 |
| parents | 625fa968d99a |
| children | 49f87ddb2c78 |
comparison
equal
deleted
inserted
replaced
| 14:625fa968d99a | 15:beb9619263b0 |
|---|---|
| 201 else { | 201 else { |
| 202 ## None supported http code error ## | 202 ## None supported http code error ## |
| 203 croak "Internal Server Error $$status..." ; | 203 croak "Internal Server Error $$status..." ; |
| 204 } | 204 } |
| 205 } | 205 } |
| 206 if ( $$status == 200 ) { print "The HMDB server returns that your request was fulfilled" ; } | 206 if ( $$status == 200 ) { print "\tThe HMDB server returns that your request (connexion test) was fulfilled\n" ; } |
| 207 } | 207 } |
| 208 | 208 |
| 209 return (1) ; | 209 return (1) ; |
| 210 } | 210 } |
| 211 ## END of SUB | 211 ## END of SUB |
| 235 | 235 |
| 236 $req->content_type('application/x-www-form-urlencoded'); | 236 $req->content_type('application/x-www-form-urlencoded'); |
| 237 $req->content('utf8=TRUE&mode='.$mode.'&query_masses='.$masses.'&tolerance='.$delta.'&database=HMDB&commit=Download Results As CSV'); | 237 $req->content('utf8=TRUE&mode='.$mode.'&query_masses='.$masses.'&tolerance='.$delta.'&database=HMDB&commit=Download Results As CSV'); |
| 238 # print Dumper $req ; | 238 # print Dumper $req ; |
| 239 my $res = $ua->request($req); | 239 my $res = $ua->request($req); |
| 240 my $status_line = undef ; | |
| 240 # print $res->as_string; | 241 # print $res->as_string; |
| 241 if ($res->is_success) { | 242 if ($res->is_success) { |
| 242 @page = split ( /\n/, $res->decoded_content ) ; | 243 @page = split ( /\n/, $res->decoded_content ) ; |
| 244 $status_line = 'OK' ; | |
| 243 } else { | 245 } else { |
| 244 my $status_line = $res->status_line ; | 246 $status_line = $res->status_line ; |
| 245 ($status_line) = ($status_line =~ /(\d+)/); | 247 ($status_line) = ($status_line =~ /(\d+)/); |
| 246 croak "HMDB service none available !! Status of the HMDB server is : $status_line\n" ; | 248 warn "\tHMDB service issue !! the server returned a $status_line HTTP error\n" ; |
| 247 } | 249 } |
| 248 | 250 |
| 249 | 251 |
| 250 return (\@page) ; | 252 return (\@page, $status_line) ; |
| 251 } | 253 } |
| 252 ## END of SUB | 254 ## END of SUB |
| 253 | 255 |
| 254 =head2 METHOD parse_hmdb_csv_results | 256 =head2 METHOD parse_hmdb_csv_results |
| 255 | 257 |
| 277 | 279 |
| 278 foreach my $line (@{$csv}) { | 280 foreach my $line (@{$csv}) { |
| 279 ## NEW HMDB format V4.0 - dec2017 | 281 ## NEW HMDB format V4.0 - dec2017 |
| 280 if ($line !~ /query_mass,compound_id,compound_name,kegg_id,formula,monoisotopic_mass,adduct,adduct_type,adduct_m\/z,delta\(ppm\)/) { | 282 if ($line !~ /query_mass,compound_id,compound_name,kegg_id,formula,monoisotopic_mass,adduct,adduct_type,adduct_m\/z,delta\(ppm\)/) { |
| 281 | 283 |
| 282 if ( $line =~ /(\d+\.\d+),(\w+),(.*),([\w|n\/a]+)\s*,(\w+),(\d+\.\d+),(.*),(\+|\-),(\d+\.\d+),(\d+)/ ) { | 284 if ( $line =~ /(\d+\.\d+),(\w+),(.*),([\w|n\/a|\s+]+)\s*,(\w+),(\d+\.\d+),(.*),(\+|\-),(\d+\.\d+),(\d+)/ ) { |
| 283 my @entry = ("$1","$2","$3","$4","$5","$6","$7","$8","$9","$10") ; | 285 my @entry = ("$1","$2","$3","$4","$5","$6","$7","$8","$9","$10") ; |
| 284 | 286 |
| 285 if ( !exists $result_by_entry{$entry[0]} ) { $result_by_entry{$entry[0]} = [] ; } | 287 if ( !exists $result_by_entry{$entry[0]} ) { $result_by_entry{$entry[0]} = [] ; } |
| 286 | 288 |
| 287 $features{ENTRY_ENTRY_ID} = $entry[1] ; | 289 $features{ENTRY_ENTRY_ID} = $entry[1] ; |
| 290 $features{ENTRY_ENTRY_NAME} = $entry[2] ; | |
| 288 $features{ENTRY_FORMULA} = $entry[4] ; | 291 $features{ENTRY_FORMULA} = $entry[4] ; |
| 289 $features{ENTRY_CPD_MZ} = $entry[5] ; | 292 $features{ENTRY_CPD_MZ} = $entry[5] ; |
| 290 $features{ENTRY_ADDUCT} = $entry[6] ; | 293 $features{ENTRY_ADDUCT} = $entry[6] ; |
| 291 $features{ENTRY_ADDUCT_TYPE} = $entry[7] ; | 294 $features{ENTRY_ADDUCT_TYPE} = $entry[7] ; |
| 292 $features{ENTRY_ADDUCT_MZ} = $entry[8] ; | 295 $features{ENTRY_ADDUCT_MZ} = $entry[8] ; |
| 293 $features{ENTRY_DELTA} = $entry[9] ; | 296 $features{ENTRY_DELTA} = $entry[9] ; |
| 294 | 297 |
| 295 my %temp = %features ; | 298 my %temp = %features ; |
| 296 push (@{$result_by_entry{$entry[0]} }, \%temp) ; | 299 push (@{$result_by_entry{$entry[0]} }, \%temp) ; |
| 297 } | 300 } |
| 298 else { | 301 # elsif ($line =~ /(\d+\.\d+)/) { # |
| 299 warn "The parsed line ($line) does not match your pattern\n " ; | 302 # ## 288.082286511284,HMDB0002255,R-Methylmalonyl-CoA, C01213 ,C25H40N7O19P3S,867.131252359,M-3H,-,288.036475,159 |
| 300 } | 303 # ## 283.108004472276,"Bicyclo_3,1,1heptane-2,3-diol,2,6,6_trimethyl","2,3-Pinanediol",n/a,C10H18O2,170.13067982,M+TFA-H,-,283.116266,29 |
| 304 # ## 174.034120330029,HMDB0011723,2-Methylhippuric acid, C01586,C10H11NO3,193.073893223,M-H20-H,-,174.055503,123 | |
| 305 # ## 193.139160745841,HMDB0012109,"7-[(1R,2R,3R,5S)-3,5-Dihydroxy-2-[(1E,3S)-3-hydroxyoct-1-en-1-yl]cyclopentyl]-5,6-dihydroxyheptanoic acid", C06475,C20H36O7,388.246103506,M-2H,-,193.115776,121 | |
| 306 # ## 214.018826827064,HMDB0011723,2-Methylhippuric acid, C01586,C10H11NO3,193.073893223,M+Na-2H,-,214.048559,139 | |
| 307 # } | |
| 308 # else { | |
| 309 # | |
| 310 # warn "The parsed line ($line) does not match your pattern\n " ; | |
| 311 # } | |
| 301 } | 312 } |
| 302 else { | 313 else { |
| 303 next ; | 314 next ; |
| 304 } | 315 } |
| 305 } ## end foreach | 316 } ## end foreach |
| 595 else { | 606 else { |
| 596 $entries->{'ENTRY_ENTRY_LOGP'} = 'NA' ; | 607 $entries->{'ENTRY_ENTRY_LOGP'} = 'NA' ; |
| 597 } | 608 } |
| 598 } | 609 } |
| 599 else { | 610 else { |
| 600 warn "This HMDB id doesn't match any collected ids\n" ; | 611 # if ($features) { |
| 612 # warn "This HMDB id doesn't match any collected ids\n" ; | |
| 613 # } | |
| 614 $entries->{'ENTRY_ENTRY_INCHI'} = 'NONEDATA' ; | |
| 615 $entries->{'ENTRY_ENTRY_LOGP'} = 'NONEDATA' ; | |
| 601 } | 616 } |
| 602 } | 617 } |
| 603 } | 618 } |
| 604 } | 619 } |
| 605 | 620 |
| 897 sub set_hmdb_matrix_object_with_ids { | 912 sub set_hmdb_matrix_object_with_ids { |
| 898 ## Retrieve Values | 913 ## Retrieve Values |
| 899 my $self = shift ; | 914 my $self = shift ; |
| 900 my ( $header, $init_mzs, $entries ) = @_ ; | 915 my ( $header, $init_mzs, $entries ) = @_ ; |
| 901 | 916 |
| 917 # print Dumper $entries ; | |
| 918 | |
| 902 my @hmdb_matrix = () ; | 919 my @hmdb_matrix = () ; |
| 903 | 920 |
| 904 if ( defined $header ) { | 921 if ( defined $header ) { |
| 905 my @headers = () ; | 922 my @headers = () ; |
| 906 | 923 |
| 932 | 949 |
| 933 if ( $check_rebond == 0 ) { | 950 if ( $check_rebond == 0 ) { |
| 934 | 951 |
| 935 push ( @anti_redondant, $entries->[$index_mz][$index_entries]{ENTRY_ENTRY_ID} ) ; | 952 push ( @anti_redondant, $entries->[$index_mz][$index_entries]{ENTRY_ENTRY_ID} ) ; |
| 936 ## | 953 ## |
| 937 my $hmdb_name = $entries->[$index_mz][$index_entries]{ENTRY_ENTRY_NAME} ; | 954 |
| 938 my $hmdb_id = $entries->[$index_mz][$index_entries]{ENTRY_ENTRY_ID} ; | 955 my ($hmdb_name, $hmdb_id, $hmdb_formula, $hmdb_cpd_mz, $hmdb_adduct, $hmdb_delta) = (undef, undef, undef, undef, undef, undef) ; |
| 939 my $hmdb_formula = $entries->[$index_mz][$index_entries]{ENTRY_FORMULA} ; | 956 |
| 940 my $hmdb_cpd_mz = $entries->[$index_mz][$index_entries]{ENTRY_CPD_MZ} ; | 957 if ($entries->[$index_mz][$index_entries]{ENTRY_ENTRY_NAME} ) { $hmdb_name = $entries->[$index_mz][$index_entries]{ENTRY_ENTRY_NAME} ; } |
| 941 my $hmdb_adduct = $entries->[$index_mz][$index_entries]{ENTRY_ADDUCT} ; | 958 else { $hmdb_name = 'UNKNOWN' ; } |
| 942 my $hmdb_delta = $entries->[$index_mz][$index_entries]{ENTRY_DELTA} ; | 959 |
| 960 if ($entries->[$index_mz][$index_entries]{ENTRY_ENTRY_ID} ) { $hmdb_id = $entries->[$index_mz][$index_entries]{ENTRY_ENTRY_ID} ; } | |
| 961 else { $hmdb_id = 0 ; } | |
| 962 | |
| 963 if ($entries->[$index_mz][$index_entries]{ENTRY_FORMULA} ) { $hmdb_formula = $entries->[$index_mz][$index_entries]{ENTRY_FORMULA} ; } | |
| 964 else { $hmdb_formula = 'N/A' ; } | |
| 965 | |
| 966 if ($entries->[$index_mz][$index_entries]{ENTRY_CPD_MZ} ) { $hmdb_cpd_mz = $entries->[$index_mz][$index_entries]{ENTRY_CPD_MZ} ; } | |
| 967 else { $hmdb_cpd_mz = 'N/A' ; } | |
| 968 | |
| 969 if ($entries->[$index_mz][$index_entries]{ENTRY_ADDUCT} ) { $hmdb_adduct = $entries->[$index_mz][$index_entries]{ENTRY_ADDUCT} ; } | |
| 970 else { $hmdb_adduct = 'N/A' ; } | |
| 971 | |
| 972 if ($entries->[$index_mz][$index_entries]{ENTRY_DELTA} ) { $hmdb_delta = $entries->[$index_mz][$index_entries]{ENTRY_DELTA} ; } | |
| 973 else { $hmdb_delta = 0 ; } | |
| 974 | |
| 943 | 975 |
| 944 ## METLIN data display model | 976 ## METLIN data display model |
| 945 ## entry1= ENTRY_DELTA::ENTRY_ENTRY_NAME::ENTRY_CPD_MZ::ENTRY_FORMULA::ENTRY_ADDUCT::ENTRY_ENTRY_ID | entry2=VAR1::VAR2::VAR3::VAR4|... | 977 ## entry1= ENTRY_DELTA::ENTRY_ENTRY_NAME::ENTRY_CPD_MZ::ENTRY_FORMULA::ENTRY_ADDUCT::ENTRY_ENTRY_ID | entry2=VAR1::VAR2::VAR3::VAR4|... |
| 946 my $entry = $hmdb_delta.'::['."$hmdb_name".']::'.$hmdb_cpd_mz.'::'.$hmdb_formula.'::['.$hmdb_adduct.']::'.$hmdb_id ; | 978 my $entry = $hmdb_delta.'::['."$hmdb_name".']::'.$hmdb_cpd_mz.'::'.$hmdb_formula.'::['.$hmdb_adduct.']::'.$hmdb_id ; |
| 947 | 979 |