Mercurial > repos > eschen42 > mqppep_anova

comparison PhosphoPeptide_Upstream_Kinase_Mapping.pl @ 6:922d309640db draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
author eschen42
date Fri, 11 Mar 2022 20:04:05 +0000
parents c1403d18c189
children d728198f1ba5
comparison
equal deleted inserted replaced
5:d4d531006735 6:922d309640db
23 # 23 #
24 # 24 #
25 ############################################################################################################################### 25 ###############################################################################################################################
26 26
27 use strict; 27 use strict;
28 use warnings; 28 #ACE use warnings;
29 use warnings 'FATAL' => 'all';
29 30
30 use Getopt::Std; 31 use Getopt::Std;
31 use DBD::SQLite::Constants qw/:file_open/; 32 use DBD::SQLite::Constants qw/:file_open/;
32 use DBI qw(:sql_types); 33 use DBI qw(:sql_types);
33 use File::Copy; 34 use File::Copy;
35 use POSIX qw(strftime); 36 use POSIX qw(strftime);
36 use Time::HiRes qw(gettimeofday); 37 use Time::HiRes qw(gettimeofday);
37 #use Data::Dump qw(dump); 38 #use Data::Dump qw(dump);
38 39
39 my $USE_SEARCH_PPEP_PY = 1; 40 my $USE_SEARCH_PPEP_PY = 1;
41 #my $FAILED_MATCH_SEQ = "Failed match";
42 my $FAILED_MATCH_SEQ = 'No Sequence';
43 my $FAILED_MATCH_GENE_NAME = 'No_Gene_Name';
40 44
41 my $dirname = dirname(__FILE__); 45 my $dirname = dirname(__FILE__);
42 my %opts; 46 my %opts;
43 my ($file_in, $average_or_sum, $db_out, $file_out, $file_melt, $phospho_type); 47 my ($file_in, $average_or_sum, $db_out, $file_out, $file_melt, $phospho_type);
44 my $dbtype; 48 my $dbtype;
45 my ($fasta_in, $networkin_in, $motifs_in, $PSP_Kinase_Substrate_in, $PSP_Regulatory_Sites_in); 49 my ($fasta_in, $networkin_in, $motifs_in, $PSP_Kinase_Substrate_in, $PSP_Regulatory_Sites_in);
46 my (@samples, %sample_id_lut, %ppep_id_lut, %data, @tmp_data, %n); 50 my (@samples, %sample_id_lut, %ppep_id_lut, %data, @tmp_data, %n);
47 my $line = 0; 51 my $line = 0;
48 my @failed_match = ("Failed match"); 52 my @failed_match = ($FAILED_MATCH_SEQ);
49 my @failed_matches; 53 my @failed_matches;
50 my (%all_data); 54 my (%all_data);
51 my (@p_peptides, @non_p_peptides); 55 my (@p_peptides, @non_p_peptides);
52 my @parsed_fasta; 56 my @parsed_fasta;
53 my (@accessions, @names, @sequences, @databases, $database); 57 my (@accessions, @names, @sequences, @databases, $database);
569 $dbtype = "SQLite"; 573 $dbtype = "SQLite";
570 } 574 }
571 575
572 print "$#accessions accessions were read from the UniProtKB/Swiss-Prot $dbtype file\n"; 576 print "$#accessions accessions were read from the UniProtKB/Swiss-Prot $dbtype file\n";
573 577
578 ######################
579 $dbh = DBI->connect("dbi:SQLite:$dbfile", undef, undef);
580 $stmth = $dbh->prepare("
581 INSERT INTO UniProtKB (
582 Uniprot_ID,
583 Description,
584 Organism_Name,
585 Organism_ID,
586 Gene_Name,
587 PE,
588 SV,
589 Sequence,
590 Database
591 ) VALUES (
592 'No Uniprot_ID',
593 'NO_GENE_SYMBOL No Description',
594 'No Organism_Name',
595 0,
596 '$FAILED_MATCH_GENE_NAME',
597 '0',
598 '0',
599 '$FAILED_MATCH_SEQ',
600 'No Database'
601 )
602 ");
603 if (not $stmth->execute()) {
604 print "Error inserting dummy row into UniProtKB: $stmth->errstr\n";
605 }
606 $dbh->disconnect if ( defined $dbh );
607 ######################
608
574 @timeData = localtime(time); 609 @timeData = localtime(time);
575 print "\n--- Start search at " . format_localtime_iso8601() ."\n"; 610 print "\n--- Start search at " . format_localtime_iso8601() ."\n";
576 611
577 print " --> Calling 'search_ppep' script\n\n"; 612 print " --> Calling 'search_ppep' script\n\n";
578 if ($verbose) { 613 if ($verbose) {
579 $i = system("\$CONDA_PREFIX/bin/python $dirname/search_ppep.py -u $db_out -p $file_in --verbose"); 614 $i = system("\$CONDA_PREFIX/bin/python $dirname/search_ppep.py -u $db_out -p $file_in --verbose");
580 } else { 615 } else {
581 $i = system("\$CONDA_PREFIX/bin/python $dirname/search_ppep.py -u $db_out -p $file_in"); 616 $i = system("\$CONDA_PREFIX/bin/python $dirname/search_ppep.py -u $db_out -p $file_in");
617 #ACE DELETEME $i = system("\$CONDA_PREFIX/bin/python $dirname/search_ppep.py -u $db_out -p $file_in --verbose");
582 } 618 }
583 if ($i) { 619 if ($i) {
584 print "python $dirname/search_ppep.py -u $db_out -p $file_in\n exited with exit code $i\n"; 620 print "python $dirname/search_ppep.py -u $db_out -p $file_in\n exited with exit code $i\n";
585 die "Search failed for phosphopeptides in SwissProt/SQLite file."; 621 die "Search failed for phosphopeptides in SwissProt/SQLite file.";
586 } 622 }
626 # AND deppep_UniProtKB.UniprotKB_ID = UniProtKB.Uniprot_ID 662 # AND deppep_UniProtKB.UniprotKB_ID = UniProtKB.Uniprot_ID
627 # ORDER BY UniprotKB_ID, deppep.seq, ppep.seq; 663 # ORDER BY UniprotKB_ID, deppep.seq, ppep.seq;
628 664
629 my %ppep_to_count_lut; 665 my %ppep_to_count_lut;
630 print "start select peptide counts " . format_localtime_iso8601() . "\n"; 666 print "start select peptide counts " . format_localtime_iso8601() . "\n";
631 $stmth = $dbh->prepare(" 667 my $uniprotkb_pep_ppep_view_stmth = $dbh->prepare("
632 SELECT DISTINCT 668 SELECT DISTINCT
633 phosphopeptide 669 phosphopeptide
634 , count(*) as i 670 , count(*) as i
635 FROM 671 FROM
636 uniprotkb_pep_ppep_view 672 uniprotkb_pep_ppep_view
637 GROUP BY 673 GROUP BY
638 phosphopeptide 674 phosphopeptide
639 ORDER BY 675 ORDER BY
640 phosphopeptide 676 phosphopeptide
641 "); 677 ");
642 if (not $stmth->execute()) { 678 if (not $uniprotkb_pep_ppep_view_stmth->execute()) {
643 die "Error fetching peptide counts: $stmth->errstr\n"; 679 die "Error fetching peptide counts: $uniprotkb_pep_ppep_view_stmth->errstr\n";
644 } 680 }
645 while (my @row = $stmth->fetchrow_array) { 681 while (my @row = $uniprotkb_pep_ppep_view_stmth->fetchrow_array) {
646 $ppep_to_count_lut{$row[0]} = $row[1]; 682 $ppep_to_count_lut{$row[0]} = $row[1];
647 #print "\$ppep_to_count_lut{$row[0]} = $ppep_to_count_lut{$row[0]}\n"; 683 #print "\$ppep_to_count_lut{$row[0]} = $ppep_to_count_lut{$row[0]}\n";
648 } 684 }
649 685
650 # accession, peptide, sequence, description, phosphopeptide, long_description, pos_start, pos_end, scrubbed, ppep_id 686 # accession, peptide, sequence, description, phosphopeptide, long_description, pos_start, pos_end, scrubbed, ppep_id
660 my $COL_SCRUBBED = 8; 696 my $COL_SCRUBBED = 8;
661 my $COL_PPEP_ID = 9; 697 my $COL_PPEP_ID = 9;
662 698
663 my %ppep_to_row_lut; 699 my %ppep_to_row_lut;
664 print "start select all records without qualification " . format_localtime_iso8601() . "\n"; 700 print "start select all records without qualification " . format_localtime_iso8601() . "\n";
665 $stmth = $dbh->prepare(" 701 $uniprotkb_pep_ppep_view_stmth = $dbh->prepare("
666 SELECT DISTINCT 702 SELECT DISTINCT
667 accession 703 accession
668 , peptide 704 , peptide
669 , sequence 705 , sequence
670 , description 706 , description
677 FROM 713 FROM
678 uniprotkb_pep_ppep_view 714 uniprotkb_pep_ppep_view
679 ORDER BY 715 ORDER BY
680 phosphopeptide 716 phosphopeptide
681 "); 717 ");
682 if (not $stmth->execute()) { 718 if (not $uniprotkb_pep_ppep_view_stmth->execute()) {
683 die "Error fetching all records without qualification: $stmth->errstr\n"; 719 die "Error fetching all records without qualification: $uniprotkb_pep_ppep_view_stmth->errstr\n";
684 } 720 }
685 my $current_ppep; 721 my $current_ppep;
686 my $counter = 0; 722 my $counter = 0;
687 my $former_ppep = ""; 723 my $former_ppep = "";
688 @tmp_matches = (); 724 @tmp_matches = ();
689 @tmp_accessions = (); 725 @tmp_accessions = ();
690 @tmp_names = (); 726 @tmp_names = ();
691 @tmp_sites = (); 727 @tmp_sites = ();
692 while (my @row = $stmth->fetchrow_array) { 728 while (my @row = $uniprotkb_pep_ppep_view_stmth->fetchrow_array) {
693 # Identify phosphopeptide for current row; 729 # Identify phosphopeptide for current row;
694 # it is an error for it to change when the counter is not zero. 730 # it is an error for it to change when the counter is not zero.
695 $current_ppep = $row[$COL_PHOSPHOPEPTIDE]; 731 $current_ppep = $row[$COL_PHOSPHOPEPTIDE];
696 732
697 # when counter is zero, prepare for a new phosphopeptide 733 # when counter is zero, prepare for a new phosphopeptide
831 867
832 my $total_length = $desired_residues_L + $tmp_p_residues[$#tmp_p_residues] - $tmp_p_residues[0] + $desired_residues_R + 1; 868 my $total_length = $desired_residues_L + $tmp_p_residues[$#tmp_p_residues] - $tmp_p_residues[0] + $desired_residues_R + 1;
833 my $arg2 = $tmp_p_residues[0] - $desired_residues_L; 869 my $arg2 = $tmp_p_residues[0] - $desired_residues_L;
834 my $arg1 = $matched_sequences{$peptide_to_match}[$i]; 870 my $arg1 = $matched_sequences{$peptide_to_match}[$i];
835 871
836 if (length($arg1) > $arg2 + $total_length - 1) { 872 if (($total_length > 0) && (length($arg1) > $arg2 + $total_length - 1)) {
873 #ACE print "\$tmp_motif = substr($arg1, $arg2, $total_length)\n";
837 $tmp_motif = substr($arg1, $arg2, $total_length); 874 $tmp_motif = substr($arg1, $arg2, $total_length);
838 #ACE print "tmp_motif = $tmp_motif\ti = $i\tpeptide_to_match = $peptide_to_match\tmatched_sequences{peptide_to_match}[i] = $matched_sequences{$peptide_to_match}[$i]\targ2 = $arg2\targ3 = $total_length\n"; 875 #ACE print "tmp_motif = $tmp_motif\ti = $i\tpeptide_to_match = $peptide_to_match\tmatched_sequences{peptide_to_match}[i] = $matched_sequences{$peptide_to_match}[$i]\targ2 = $arg2\targ3 = $total_length\n";
839 876
840 # Put the "p" back in front of the appropriate phospho-residue(s). 877 # Put the "p" back in front of the appropriate phospho-residue(s).
841 my (@tmp_residues, $tmp_position); 878 my (@tmp_residues, $tmp_position);
1662 1699
1663 ($start_seconds, $start_microseconds) = gettimeofday; 1700 ($start_seconds, $start_microseconds) = gettimeofday;
1664 1701
1665 print "Writing PSP_Regulatory_site records\n"; 1702 print "Writing PSP_Regulatory_site records\n";
1666 1703
1667 #ACE $stmth = $dbh->prepare(" 1704 my $psp_regulatory_site_stmth = $dbh->prepare("
1668 #ACE INSERT INTO PSP_Regulatory_site (
1669 #ACE DOMAIN,
1670 #ACE ON_FUNCTION,
1671 #ACE ON_PROCESS,
1672 #ACE ON_PROT_INTERACT,
1673 #ACE ON_OTHER_INTERACT,
1674 #ACE NOTES,
1675 #ACE SITE_PLUSMINUS_7AA,
1676 #ACE ORGANISM,
1677 #ACE PROTEIN
1678 #ACE ) VALUES (?,?,?,?,?,?,?,?,?)
1679 #ACE ");
1680
1681 $stmth = $dbh->prepare("
1682 INSERT INTO PSP_Regulatory_site ( 1705 INSERT INTO PSP_Regulatory_site (
1683 DOMAIN, 1706 DOMAIN,
1684 ON_FUNCTION, 1707 ON_FUNCTION,
1685 ON_PROCESS, 1708 ON_PROCESS,
1686 ON_PROT_INTERACT, 1709 ON_PROT_INTERACT,
1692 "); 1715 ");
1693 1716
1694 foreach my $peptide (keys %data) { 1717 foreach my $peptide (keys %data) {
1695 if (exists($domain_2{$peptide}) and (defined $domain_2{$peptide}) and (not $domain_2{$peptide} eq "") ) { 1718 if (exists($domain_2{$peptide}) and (defined $domain_2{$peptide}) and (not $domain_2{$peptide} eq "") ) {
1696 #ACE print "writing domain $domain_2{$peptide} for regulatory site(s) $seq_plus7aa_2{$peptide}\n"; #ACE 1719 #ACE print "writing domain $domain_2{$peptide} for regulatory site(s) $seq_plus7aa_2{$peptide}\n"; #ACE
1697 $stmth->bind_param(1, $domain_2{$peptide}); 1720 $psp_regulatory_site_stmth->bind_param(1, $domain_2{$peptide});
1698 $stmth->bind_param(2, $ON_FUNCTION_2{$peptide}); 1721 $psp_regulatory_site_stmth->bind_param(2, $ON_FUNCTION_2{$peptide});
1699 $stmth->bind_param(3, $ON_PROCESS_2{$peptide}); 1722 $psp_regulatory_site_stmth->bind_param(3, $ON_PROCESS_2{$peptide});
1700 $stmth->bind_param(4, $ON_PROT_INTERACT_2{$peptide}); 1723 $psp_regulatory_site_stmth->bind_param(4, $ON_PROT_INTERACT_2{$peptide});
1701 $stmth->bind_param(5, $ON_OTHER_INTERACT_2{$peptide}); 1724 $psp_regulatory_site_stmth->bind_param(5, $ON_OTHER_INTERACT_2{$peptide});
1702 $stmth->bind_param(6, $notes_2{$peptide}); 1725 $psp_regulatory_site_stmth->bind_param(6, $notes_2{$peptide});
1703 $stmth->bind_param(7, $seq_plus7aa_2{$peptide}); 1726 $psp_regulatory_site_stmth->bind_param(7, $seq_plus7aa_2{$peptide});
1704 $stmth->bind_param(8, $organism_2{$peptide}); 1727 $psp_regulatory_site_stmth->bind_param(8, $organism_2{$peptide});
1705 #ACE $stmth->bind_param(9, $psp_regsite_protein_2{$peptide}); 1728 if (not $psp_regulatory_site_stmth->execute()) {
1706 if (not $stmth->execute()) { 1729 print "Error writing PSP_Regulatory_site for one regulatory site with peptide '$domain_2{$peptide}': $psp_regulatory_site_stmth->errstr\n";
1707 print "Error writing PSP_Regulatory_site for one regulatory site with peptide '$domain_2{$peptide}': $stmth->errstr\n";
1708 } else { 1730 } else {
1709 #ACE print "added domain for $domain_2{$peptide}\n"; 1731 #ACE print "added domain for $domain_2{$peptide}\n";
1710 } 1732 }
1711 } elsif (exists($domain_2{$peptide}) and (not defined $domain_2{$peptide})) { 1733 } elsif (exists($domain_2{$peptide}) and (not defined $domain_2{$peptide})) {
1712 print "\$domain_2{$peptide} is undefined\n"; #ACE 1734 print "\$domain_2{$peptide} is undefined\n"; #ACE
1713 } 1735 }
1714 } 1736 }
1715 1737
1716 $dbh->{AutoCommit} = $auto_commit; 1738 $dbh->{AutoCommit} = $auto_commit;
1717 # auto_commit implicitly finishes stmth, apparently # $stmth->finish; 1739 # auto_commit implicitly finishes psp_regulatory_site_stmth, apparently # $psp_regulatory_site_stmth->finish;
1718 $dbh->disconnect if ( defined $dbh ); 1740 $dbh->disconnect if ( defined $dbh );
1719 1741
1720 1742
1721 ($end_seconds, $end_microseconds) = gettimeofday; 1743 ($end_seconds, $end_microseconds) = gettimeofday;
1722 1744
1870 # column 1: p-peptide 1892 # column 1: p-peptide
1871 print OUT "$peptide\t"; 1893 print OUT "$peptide\t";
1872 push (@ppep_metadata, $ppep_id); 1894 push (@ppep_metadata, $ppep_id);
1873 push (@ppep_intensity, $peptide); 1895 push (@ppep_intensity, $peptide);
1874 1896
1897 my $verbose_cond = 0; # $peptide eq 'AAAAAAAGDpSDpSWDADAFSVEDPVR' || $peptide eq 'KKGGpSpSDEGPEPEAEEpSDLDSGSVHSASGRPDGPVR';
1875 # skip over failed matches 1898 # skip over failed matches
1876 if ($matched_sequences{$peptide} eq "Failed match") { 1899 print "\nfirst match for '$peptide' is '$matched_sequences{$peptide}[0]' and FAILED_MATCH_SEQ is '$FAILED_MATCH_SEQ'\n" if $verbose_cond;
1900 if ($matched_sequences{$peptide}[0] eq $FAILED_MATCH_SEQ) {
1901 # column 2: Protein description
1902 # column 3: Gene name(s)
1903 # column 4: FASTA name
1904 # column 5: phospho-residues
1905 # Column 6: UNIQUE phospho-motifs
1906 # Column 7: accessions
1907 # Column 8: ALL motifs with residue numbers
1908 # 2 3 4 5 6 7 8
1877 print OUT "Sequence not found in FASTA database\tNA\tNA\tNA\tNA\tNA\tNA\t"; 1909 print OUT "Sequence not found in FASTA database\tNA\tNA\tNA\tNA\tNA\tNA\t";
1910 print "No match found for '$peptide' in sequence database\n";
1911 $gene_names = '$FAILED_MATCH_GENE_NAME';
1878 } else { 1912 } else {
1879 my @description = (); 1913 my @description = ();
1880 my %seen = (); 1914 my %seen = ();
1881 # Print just the protein description 1915 # Print just the protein description
1882 for $i (0 .. $#{$names{$peptide}}) { 1916 for $i (0 .. $#{$names{$peptide}}) {
1902 # column 3: Gene name(s) 1936 # column 3: Gene name(s)
1903 $gene_names = join(" /// ", @gene); 1937 $gene_names = join(" /// ", @gene);
1904 print OUT $gene_names, "\t"; 1938 print OUT $gene_names, "\t";
1905 push (@ppep_metadata, join(" /// ", @gene)); 1939 push (@ppep_metadata, join(" /// ", @gene));
1906 1940
1907 # print the entire names
1908 # column 4: FASTA name 1941 # column 4: FASTA name
1909 print OUT join(" /// ", @{$names{$peptide}}), "\t"; 1942 print OUT join(" /// ", @{$names{$peptide}}), "\t";
1910 push (@ppep_metadata, join(" /// ", @{$names{$peptide}})); 1943 push (@ppep_metadata, join(" /// ", @{$names{$peptide}}));
1911 1944
1912 # Print the phospho-residues 1945 # column 5: phospho-residues
1913 # column 5:
1914 my $tmp_for_insert = ""; 1946 my $tmp_for_insert = "";
1947 my $foobar;
1915 for my $i (0 .. $#{ $matched_sequences{$peptide} } ) { 1948 for my $i (0 .. $#{ $matched_sequences{$peptide} } ) {
1949 print "match $i for '$peptide' is '$matched_sequences{$peptide}[$i]'\n" if $verbose_cond;
1916 if ($i < $#{ $matched_sequences{$peptide} }) { 1950 if ($i < $#{ $matched_sequences{$peptide} }) {
1917 if (defined $p_residues{$peptide}{$i}) { 1951 if (defined $p_residues{$peptide}{$i}) {
1918 @tmp_p_residues = @{$p_residues{$peptide}{$i}}; 1952 @tmp_p_residues = @{$p_residues{$peptide}{$i}};
1919 for $j (0 .. $#tmp_p_residues) { 1953 for $j (0 .. $#tmp_p_residues) {
1920 if ($j < $#tmp_p_residues) { 1954 if ($j < $#tmp_p_residues) {
1929 } 1963 }
1930 } 1964 }
1931 } 1965 }
1932 } 1966 }
1933 elsif ($i == $#{ $matched_sequences{$peptide} }) { 1967 elsif ($i == $#{ $matched_sequences{$peptide} }) {
1968 my $there_were_sites = 0;
1934 if (defined $p_residues{$peptide}{$i}) { 1969 if (defined $p_residues{$peptide}{$i}) {
1935 @tmp_p_residues = @{$p_residues{$peptide}{$i}}; 1970 @tmp_p_residues = @{$p_residues{$peptide}{$i}};
1936 for my $j (0 .. $#tmp_p_residues) { 1971 if ($#tmp_p_residues > 0) {
1937 if ($j < $#tmp_p_residues) { 1972 for my $j (0 .. $#tmp_p_residues) {
1938 my $tmp_site_for_printing = $p_residues{$peptide}{$i}[$j] + 1; # added 12.05.2012 for Justin's data 1973 if ($j < $#tmp_p_residues) {
1939 print OUT "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing, "; 1974 if (defined $p_residues{$peptide}{$i}[$j]) {
1940 $tmp_for_insert .= "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing, "; 1975 my $tmp_site_for_printing = $p_residues{$peptide}{$i}[$j] + 1; # added 12.05.2012 for Justin's data
1976 $foobar = $residues{$peptide}{$i}[$j];
1977 if (defined $foobar) {
1978 print OUT "$foobar";
1979 print OUT "$tmp_site_for_printing, ";
1980 $tmp_for_insert .= "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing, ";
1981 $there_were_sites = 1;
1982 }
1983 }
1984 }
1985 elsif ($j == $#tmp_p_residues) {
1986 if (defined $p_residues{$peptide}{$i}[$j]) {
1987 $foobar = $residues{$peptide}{$i}[$j];
1988 if (defined $foobar) {
1989 my $tmp_site_for_printing = $p_residues{$peptide}{$i}[$j] + 1; # added 12.05.2012 for Justin's data
1990 print OUT "$foobar";
1991 print OUT "$tmp_site_for_printing\t";
1992 #ACE print OUT "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing\t";
1993 $tmp_for_insert .= "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing";
1994 $there_were_sites = 1;
1995 }
1996 }
1997 }
1941 } 1998 }
1942 elsif ($j == $#tmp_p_residues) { 1999 }
1943 my $tmp_site_for_printing = $p_residues{$peptide}{$i}[$j] + 1; # added 12.05.2012 for Justin's data 2000 }
1944 print OUT "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing\t"; 2001 if (0 == $there_were_sites) {
1945 $tmp_for_insert .= "p$residues{$peptide}{$i}[$j]$tmp_site_for_printing";
1946 }
1947 }
1948 } else {
1949 print OUT "\t"; 2002 print OUT "\t";
1950 } 2003 }
1951 } 2004 }
1952 } 2005 }
2006 print "tmp_for_insert '$tmp_for_insert' for '$peptide'\n" if $verbose_cond;
1953 push (@ppep_metadata, $tmp_for_insert); 2007 push (@ppep_metadata, $tmp_for_insert);
1954 2008
1955 # Print the UNIQUE phospho-motifs 2009 # Column 6: UNIQUE phospho-motifs
1956 # Column 6:
1957 print OUT join(" /// ", @{$unique_motifs{$peptide}}), "\t"; 2010 print OUT join(" /// ", @{$unique_motifs{$peptide}}), "\t";
1958 push (@ppep_metadata, join(" /// ", @{$unique_motifs{$peptide}})); 2011 push (@ppep_metadata, join(" /// ", @{$unique_motifs{$peptide}}));
1959 2012
1960 # Print the accessions 2013 # Column 7: accessions
1961 # Column 7:
1962 if (defined $accessions{$peptide}) { 2014 if (defined $accessions{$peptide}) {
1963 print OUT join(" /// ", @{$accessions{$peptide}}), "\t"; 2015 print OUT join(" /// ", @{$accessions{$peptide}}), "\t";
1964 push (@ppep_metadata, join(" /// ", @{$accessions{$peptide}})); 2016 push (@ppep_metadata, join(" /// ", @{$accessions{$peptide}}));
1965 } else { 2017 } else {
1966 print OUT "\t"; 2018 print OUT "\t";
1967 push (@ppep_metadata, ""); 2019 push (@ppep_metadata, "");
1968 } 2020 }
1969 2021
1970 # print ALL motifs with residue numbers 2022 # Column 8: ALL motifs with residue numbers
1971 # Column 8:
1972 if (defined $p_motifs{$peptide}) { 2023 if (defined $p_motifs{$peptide}) {
1973 print OUT join(" /// ", @{$p_motifs{$peptide}}), "\t"; 2024 print OUT join(" /// ", @{$p_motifs{$peptide}}), "\t";
1974 push (@ppep_metadata, join(" /// ", @{$p_motifs{$peptide}})); 2025 push (@ppep_metadata, join(" /// ", @{$p_motifs{$peptide}}));
1975 } else { 2026 } else {
1976 print OUT "\t"; 2027 print OUT "\t";