libpappsomspp
Library for mass spectrometry
Loading...
Searching...
No Matches
Classes | Public Types | Public Member Functions | Private Member Functions | Private Attributes | List of all members
pappso::cbor::psm::SageTsvHandler Class Reference

#include <sagetsvhandler.h>

Inheritance diagram for pappso::cbor::psm::SageTsvHandler:

Classes

struct  Line
 
struct  Psm
 
struct  Sample
 
struct  Scan
 

Public Types

enum class  Columns : std::int8_t {
  psm_id , peptide , proteins , num_proteins ,
  filename , scannr , rank , label ,
  expmass , calcmass , charge , peptide_len ,
  missed_cleavages , semi_enzymatic , isotope_error , precursor_ppm ,
  fragment_ppm , hyperscore , delta_next , delta_best ,
  rt , aligned_rt , predicted_rt , delta_rt_model ,
  ion_mobility , predicted_mobility , delta_mobility , matched_peaks ,
  longest_b , longest_y , longest_y_pct , matched_intensity_pct ,
  scored_candidates , poisson , sage_discriminant_score , posterior_error ,
  spectrum_q , peptide_q , protein_q , ms2_intensity
}
 

Public Member Functions

 SageTsvHandler (pappso::UiMonitorInterface *p_monitor, const SageReader &sage_reader, bool is_fasta_pass, PsmProteinMap &psm_protein_map)
 
virtual ~SageTsvHandler ()
 
virtual void startSheet (const QString &sheet_name) override
 
virtual void endSheet () override
 
virtual void startLine () override
 
virtual void endLine () override
 
virtual void setCell (const OdsCell &cell) override
 
virtual void endDocument () override
 

Private Member Functions

void parsePeptide (const QString &peptide_str)
 
void parseProteins (const QString &proteins_str)
 
bool parseSpectrumStringId (const QString &spectrum_string_id)
 
void parseMsRunFilename (const QString &msrun_filename)
 
void recordLine ()
 
void writeSample (const Sample &one_sample)
 
void writeScan (const Scan &one_scan)
 
void writePsm (const Psm &one_psm)
 

Private Attributes

const SageReader & m_sageReader
 
PsmProteinMap & m_psmProteinMap
 
pappso::UiMonitorInterface * mp_monitor = nullptr
 
std::size_t m_progressIndex = 0
 
std::size_t m_lineNumber = 0
 
std::size_t m_columnNumber = 0
 
std::vector< Columns > m_columnTypeList
 
pappso::PeptideSp msp_peptide
 
std::vector< SageReader::SageModification > m_staticModificationList
 
std::vector< SageReader::SageModification > m_variableModificationList
 
std::size_t m_spectrumIndex
 
Line m_line
 
QString m_decoyTag
 
bool m_isFirstPass
 
QStringList m_proteinList
 
QString m_spectrumNativeId
 
std::map< QString, Sample > m_sampleMap
 
Sample * mp_currentSample
 

Detailed Description

Todo
write docs

Definition at line 49 of file sagetsvhandler.h.

Member Enumeration Documentation

◆ Columns

enum class pappso::cbor::psm::SageTsvHandler::Columns : std::int8_t
strong
Enumerator
psm_id 
peptide 
proteins 
num_proteins 
filename 
scannr 
rank 
label 
expmass 
calcmass 
charge 
peptide_len 
missed_cleavages 
semi_enzymatic 
isotope_error 
precursor_ppm 
fragment_ppm 
hyperscore 
delta_next 
delta_best 
rt 
aligned_rt 
predicted_rt 
delta_rt_model 
ion_mobility 
predicted_mobility 
delta_mobility 
matched_peaks 
longest_b 
longest_y 
longest_y_pct 
matched_intensity_pct 
scored_candidates 
poisson 
sage_discriminant_score 
posterior_error 
spectrum_q 
peptide_q 
protein_q 
ms2_intensity 

Definition at line 52 of file sagetsvhandler.h.

53 {
54 psm_id,
55 peptide,
56 proteins,
57 num_proteins,
58 filename,
59 scannr,
60 rank,
61 label,
62 expmass,
63 calcmass,
64 charge,
65 peptide_len,
66 missed_cleavages,
67 semi_enzymatic,
68 isotope_error,
69 precursor_ppm,
70 fragment_ppm,
71 hyperscore,
72 delta_next,
73 delta_best,
74 rt,
75 aligned_rt,
76 predicted_rt,
77 delta_rt_model,
78 ion_mobility,
79 predicted_mobility,
80 delta_mobility,
81 matched_peaks,
82 longest_b,
83 longest_y,
84 longest_y_pct,
85 matched_intensity_pct,
86 scored_candidates,
87 poisson,
88 sage_discriminant_score,
89 posterior_error,
90 spectrum_q,
91 peptide_q,
92 protein_q,
93 ms2_intensity,
94 };
pappso::Enums::DataKind::rt
@ rt
Retention time.
Definition types.h:251

Constructor & Destructor Documentation

◆ SageTsvHandler()

pappso::cbor::psm::SageTsvHandler::SageTsvHandler ( pappso::UiMonitorInterface * p_monitor,
const SageReader & sage_reader,
bool is_fasta_pass,
PsmProteinMap & psm_protein_map )

Default constructor

Definition at line 47 of file sagetsvhandler.cpp.

51 : m_sageReader(sage_reader), m_psmProteinMap(psm_protein_map)
52{
53 mp_monitor = p_monitor;
54 m_staticModificationList = sage_reader.getStaticModificationList();
55 m_variableModificationList = sage_reader.getVariableModificationList();
56 m_decoyTag = sage_reader.getDecoyTag();
57 m_isFirstPass = is_fasta_pass;
58}
pappso::cbor::psm::SageTsvHandler::m_staticModificationList
std::vector< SageReader::SageModification > m_staticModificationList
Definition sagetsvhandler.h:225
pappso::cbor::psm::SageTsvHandler::mp_monitor
pappso::UiMonitorInterface * mp_monitor
Definition sagetsvhandler.h:218
pappso::cbor::psm::SageTsvHandler::m_variableModificationList
std::vector< SageReader::SageModification > m_variableModificationList
Definition sagetsvhandler.h:226

References pappso::cbor::psm::SageReader::getDecoyTag(), pappso::cbor::psm::SageReader::getStaticModificationList(), pappso::cbor::psm::SageReader::getVariableModificationList(), m_decoyTag, m_isFirstPass, m_psmProteinMap, m_sageReader, m_staticModificationList, m_variableModificationList, and mp_monitor.

◆ ~SageTsvHandler()

pappso::cbor::psm::SageTsvHandler::~SageTsvHandler ( )
virtual

Destructor

Definition at line 60 of file sagetsvhandler.cpp.

61{
62}

Member Function Documentation

◆ endDocument()

void pappso::cbor::psm::SageTsvHandler::endDocument ( )
overridevirtual

callback that report the end of the ODS document. Override it if you need to know that reading is finished.

Definition at line 65 of file sagetsvhandler.cpp.

66{
67 for(auto it_sample_map : m_sampleMap)
68 {
69 writeSample(it_sample_map.second);
70 }
71}
pappso::cbor::psm::SageTsvHandler::writeSample
void writeSample(const Sample &one_sample)
Definition sagetsvhandler.cpp:723
pappso::cbor::psm::SageTsvHandler::m_sampleMap
std::map< QString, Sample > m_sampleMap
Definition sagetsvhandler.h:233

References m_sampleMap, and writeSample().

◆ endLine()

void pappso::cbor::psm::SageTsvHandler::endLine ( )
overridevirtual

callback that indicates a line ending. Override it if needed.

Definition at line 74 of file sagetsvhandler.cpp.

75{
76 if(m_line.charge != 0)
77 recordLine();
78 m_lineNumber++;
79}

References m_line, m_lineNumber, and recordLine().

◆ endSheet()

void pappso::cbor::psm::SageTsvHandler::endSheet ( )
overridevirtual

callback that indicates the end of the current data sheet. Override it if needed

Definition at line 82 of file sagetsvhandler.cpp.

83{
84}

◆ parseMsRunFilename()

void pappso::cbor::psm::SageTsvHandler::parseMsRunFilename ( const QString & msrun_filename)
private

Definition at line 541 of file sagetsvhandler.cpp.

542{
543
544 if(m_isFirstPass)
545 return;
546
547 // find the sample :
548 auto it_insert = m_sampleMap.insert({msrun_filename, {}});
549 mp_currentSample = &(it_insert.first->second);
550 if(it_insert.second)
551 {
552 // new sample
553 it_insert.first->second.name = msrun_filename;
554 QCborMap ms_file;
555 QCborMap identification_file;
556
557 it_insert.first->second.cbor_core_sample.insert(QString("name"),
558 QFileInfo(msrun_filename).baseName());
559
560 // identification_file_list
561 QCborArray identification_file_list;
562 identification_file.insert(QString("name"), m_sageReader.getmJsonAbsoluteFilePath());
563 identification_file_list.push_back(identification_file);
564 it_insert.first->second.cbor_core_sample.insert(QString("identification_file_list"),
565 identification_file_list);
566
567
568 ms_file.insert(QString("name"), QFileInfo(msrun_filename).absoluteFilePath());
569 it_insert.first->second.cbor_core_sample.insert(QString("peaklist_file"), ms_file);
570 }
571
572 /*
573 msp_msrun = m_sageReader.getSageFileReader().getMsRunSpWithFileName(msrun_filename);
574 qDebug() << msp_msrun.get()->getFileName();
575
576 msp_identificationSageJsonFileSp =
577 m_sageReader.getSageFileReader().getIdentificationSageJsonFileSpWithFileName(msrun_filename);
578
579 mp_identificationGroup =
580 m_sageReader.getSageFileReader().getIdentificationGroupPtrWithFileName(msrun_filename);
581 qDebug() << msp_msrun.get()->getFileName();
582 */
583}

References m_isFirstPass, m_sageReader, m_sampleMap, and mp_currentSample.

Referenced by setCell().

◆ parsePeptide()

void pappso::cbor::psm::SageTsvHandler::parsePeptide ( const QString & peptide_str)
private

Definition at line 438 of file sagetsvhandler.cpp.

439{
440 qDebug();
441 if(m_isFirstPass)
442 return;
443 QString peptide_str_verif = peptide_str;
444 // fixed modifications :
445 for(SageReader::SageModification modif : m_staticModificationList)
446 {
447 qDebug() << modif.strModification;
448 qDebug() << modif.modification->getAccession();
449 peptide_str_verif = peptide_str_verif.replace(
450 modif.strModification, QString("[%1]").arg(modif.modification->getAccession()));
451 }
452 // variable modifications :
453 for(SageReader::SageModification modif : m_variableModificationList)
454 {
455 qDebug() << modif.strModification;
456 qDebug() << modif.modification->getAccession();
457 peptide_str_verif = peptide_str_verif.replace(
458 modif.strModification, QString("[%1]").arg(modif.modification->getAccession()));
459 }
460
461 qDebug() << peptide_str_verif;
462 // LPMFGC[+57.0216]NDATQVYK
463 msp_peptide = pappso::PeptideProFormaParser::parseString(peptide_str_verif);
464 qDebug();
465 // variable modifications :
466 /*
467 setVariableModifications(peptide_sp,
468 peptide_line.peptide_string_list.at(6));
469*/
470 qDebug() << msp_peptide.get()->toProForma();
471}
pappso::PeptideProFormaParser::parseString
static PeptideSp parseString(const QString &pepstr)
Definition peptideproformaparser.cpp:227

References m_isFirstPass, m_staticModificationList, m_variableModificationList, msp_peptide, and pappso::PeptideProFormaParser::parseString().

Referenced by setCell().

◆ parseProteins()

void pappso::cbor::psm::SageTsvHandler::parseProteins ( const QString & proteins_str)
private

Definition at line 475 of file sagetsvhandler.cpp.

476{
477 m_proteinList.clear();
478 m_proteinList = proteins_str.split(";");
479 if(m_isFirstPass)
480 {
481 for(QString accession : m_proteinList)
482 {
483
484 PsmProtein psm_protein;
485 psm_protein.protein_sp = std::make_shared<pappso::Protein>(accession, "");
486 psm_protein.isTarget = true;
487 if(accession.startsWith(m_decoyTag))
488 {
489 psm_protein.isTarget = false;
490 }
491
492 m_psmProteinMap.insert(psm_protein);
493 }
494 }
495}

References pappso::cbor::psm::PsmProtein::isTarget, m_decoyTag, m_isFirstPass, m_proteinList, m_psmProteinMap, and pappso::cbor::psm::PsmProtein::protein_sp.

Referenced by setCell().

◆ parseSpectrumStringId()

bool pappso::cbor::psm::SageTsvHandler::parseSpectrumStringId ( const QString & spectrum_string_id)
private

Definition at line 498 of file sagetsvhandler.cpp.

499{
500 m_spectrumNativeId = spectrum_string_id;
501 qDebug() << spectrum_string_id;
502 if(m_isFirstPass)
503 return false;
504 // controllerType=0 controllerNumber=1 scan=176056
505
506 bool is_ok = false;
507 QStringList scan_list = spectrum_string_id.split("scan=");
508 if(scan_list.size() == 2)
509 {
510 // we bet that there is a scan number, easy to parse
511 m_spectrumIndex = scan_list.at(1).toULongLong(&is_ok);
512 if(m_spectrumIndex > 0)
513 m_spectrumIndex--;
514 }
515 if(is_ok == false)
516 { /*
517 if(msp_previousMsrun != msp_msrun)
518 {
519 mp_monitor->setStatus(
520 QObject::tr("Reading mz data file %1").arg(msp_msrun.get()->getFileName()));
521 msp_previousMsrun = msp_msrun;
522 }
523
524 pappso::MsRunReader *msrunreader_p = msp_msrun.get()->getMsRunReaderSPtr().get();
525 if(msrunreader_p->getMsRunId()->getMsDataFormat() == pappso::MsDataFormat::brukerTims)
526 {
527 m_spectrumIndex = spectrum_string_id.toInt() * 2 - 1;
528 }
529 else
530 {
531 m_spectrumIndex =
532 msrunreader_p->spectrumStringIdentifier2SpectrumIndex(spectrum_string_id);
533 }*/
534 }
535
536 qDebug() << spectrum_string_id;
537 return is_ok;
538}

References m_isFirstPass, m_spectrumIndex, and m_spectrumNativeId.

Referenced by setCell().

◆ recordLine()

void pappso::cbor::psm::SageTsvHandler::recordLine ( )
private

Definition at line 586 of file sagetsvhandler.cpp.

587{
588 qDebug();
589
590 if(m_isFirstPass)
591 {
592 for(const QString &accession : m_proteinList)
593 {
594 PsmProtein psm_protein;
595 psm_protein.protein_sp = std::make_shared<pappso::Protein>(accession, "");
596
597 auto it = m_psmProteinMap.insert(psm_protein);
598 it.first->second.cborEval.insert(QString("protein_q"), m_line.protein_q);
599 }
600 }
601 else
602 {
603 /*
604 PeptideEvidence pe(msp_msrun.get(), m_spectrumIndex, true);
605 pe.setCharge(m_line.charge);
606 pe.setChecked(true);
607 pe.setExperimentalMass(m_line.expmass);
608 pe.setPeptideXtpSp(msp_peptide);
609 pe.setIdentificationDataSource(msp_identificationSageJsonFileSp.get());
610 pe.setIdentificationEngine(m_identificationEngine);
611 pe.setRetentionTime(m_line.rt);
612 pe.setParam(PeptideEvidenceParam::tandem_hyperscore, m_line.hyperscore);
613 pe.setParam(PeptideEvidenceParam::sage_sage_discriminant_score,
614 m_line.sage_discriminant_score);
615 pe.setParam(PeptideEvidenceParam::sage_peptide_q, m_line.peptide_q);
616 pe.setParam(PeptideEvidenceParam::sage_posterior_error, m_line.posterior_error);
617 pe.setParam(PeptideEvidenceParam::sage_spectrum_q, m_line.spectrum_q);
618 pe.setParam(PeptideEvidenceParam::sage_predicted_rt, m_line.predicted_rt);
619 pe.setParam(PeptideEvidenceParam::sage_isotope_error, m_line.isotope_error);
620
621
622 PeptideMatch peptide_match;
623 // peptide_match.setStart(mz_peptide_evidence.start);
624 peptide_match.setPeptideEvidenceSp(
625 msp_identificationSageJsonFileSp.get()->getPeptideEvidenceStore().getInstance(&pe));
626 */
627
628 // find the scan in sample
629 auto it_insert = mp_currentSample->scan_map.insert({m_spectrumNativeId, Scan()});
630 Scan *current_cbor_scan_p = &(it_insert.first->second);
631 if(it_insert.second)
632 {
633 // new scan
634 QCborMap &scan_id = it_insert.first->second.cbor_id;
635 scan_id.insert(QString("index"), (qint64)m_spectrumIndex);
636 scan_id.insert(QString("native_id"), m_spectrumNativeId);
637
638 QCborMap &scan_ms2 = it_insert.first->second.cbor_ms2;
639 scan_ms2.insert(QString("rt"), m_line.rt);
640
641 QCborMap &scan_precursor = it_insert.first->second.cbor_precursor;
642 scan_precursor.insert(QString("z"), m_line.charge);
643 scan_precursor.insert(QString("mz"), m_line.expmass);
644 }
645
646 Psm one_psm;
647 one_psm.proforma = msp_peptide.get()->toProForma();
648
649 for(const QString &accession : m_proteinList)
650 {
651 QCborMap cbor_protein;
652 cbor_protein.insert(QString("accession"), accession);
653
654
655 // start/end positions
656 QString protein_sequence =
657 QString(m_psmProteinMap.getByAccession(accession).protein_sp.get()->getSequence())
658 .replace("L", "I");
659 QString peptide_sequence = msp_peptide.get()->getSequenceLi();
660 int position = protein_sequence.indexOf(peptide_sequence);
661
662 QCborArray positions;
663 while(position >= 0)
664 {
665 positions.push_back(position);
666 position = protein_sequence.indexOf(peptide_sequence, position + 1);
667 }
668
669 cbor_protein.insert(QString("positions"), positions);
670
671 one_psm.cbor_protein_list.append(cbor_protein);
672 }
673
674 one_psm.cbor_eval.insert(QString("aligned_rt"), m_line.aligned_rt);
675 one_psm.cbor_eval.insert(QString("calcmass"), m_line.calcmass);
676 one_psm.cbor_eval.insert(QString("delta_best"), m_line.delta_best);
677 one_psm.cbor_eval.insert(QString("delta_mobility"), m_line.delta_mobility);
678 one_psm.cbor_eval.insert(QString("delta_next"), m_line.delta_next);
679 one_psm.cbor_eval.insert(QString("delta_rt_model"), m_line.delta_rt_model);
680 one_psm.cbor_eval.insert(QString("fragment_ppm"), m_line.fragment_ppm);
681 one_psm.cbor_eval.insert(QString("hyperscore"), m_line.hyperscore);
682 one_psm.cbor_eval.insert(QString("ion_mobility"), m_line.ion_mobility);
683 one_psm.cbor_eval.insert(QString("isotope_error"), m_line.isotope_error);
684 one_psm.cbor_eval.insert(QString("label"), m_line.label);
685 one_psm.cbor_eval.insert(QString("longest_b"), (qint64)m_line.longest_b);
686 one_psm.cbor_eval.insert(QString("longest_y"), (qint64)m_line.longest_y);
687 one_psm.cbor_eval.insert(QString("longest_y_pct"), m_line.longest_y_pct);
688 one_psm.cbor_eval.insert(QString("matched_intensity_pct"), m_line.matched_intensity_pct);
689 one_psm.cbor_eval.insert(QString("matched_peaks"), (qint64)m_line.matched_peaks);
690 one_psm.cbor_eval.insert(QString("missed_cleavages"), m_line.missed_cleavages);
691 one_psm.cbor_eval.insert(QString("ms2_intensity"), m_line.ms2_intensity);
692 one_psm.cbor_eval.insert(QString("peptide_len"), (qint64)m_line.peptide_len);
693 one_psm.cbor_eval.insert(QString("peptide_q"), m_line.peptide_q);
694 one_psm.cbor_eval.insert(QString("poisson"), m_line.poisson);
695 one_psm.cbor_eval.insert(QString("posterior_error"), m_line.posterior_error);
696 one_psm.cbor_eval.insert(QString("precursor_ppm"), m_line.precursor_ppm);
697 one_psm.cbor_eval.insert(QString("predicted_mobility"), m_line.predicted_mobility);
698 one_psm.cbor_eval.insert(QString("predicted_rt"), m_line.predicted_rt);
699 one_psm.cbor_eval.insert(QString("protein_q"), m_line.protein_q);
700 one_psm.cbor_eval.insert(QString("rank"), m_line.rank);
701 one_psm.cbor_eval.insert(QString("sage_discriminant_score"), m_line.sage_discriminant_score);
702 one_psm.cbor_eval.insert(QString("scored_candidates"), (qint64)m_line.scored_candidates);
703 one_psm.cbor_eval.insert(QString("semi_enzymatic"), m_line.semi_enzymatic);
704 one_psm.cbor_eval.insert(QString("spectrum_q"), m_line.spectrum_q);
705
706 current_cbor_scan_p->psm_list.emplace_back(one_psm);
707 }
708
709 std::size_t progress = m_lineNumber / 10000;
710 if(progress > m_progressIndex)
711 {
712 if(mp_monitor->shouldIstop())
713 {
714 throw pappso::ExceptionInterrupted(
715 QObject::tr("Sage TSV data reading process interrupted"));
716 }
717 m_progressIndex = progress;
718 mp_monitor->setStatus(QString("%1K ").arg(m_progressIndex * 10));
719 }
720}

References pappso::cbor::psm::SageTsvHandler::Psm::cbor_eval, pappso::cbor::psm::SageTsvHandler::Psm::cbor_protein_list, m_isFirstPass, m_line, m_lineNumber, m_progressIndex, m_proteinList, m_psmProteinMap, m_spectrumIndex, m_spectrumNativeId, mp_currentSample, mp_monitor, msp_peptide, pappso::cbor::psm::SageTsvHandler::Psm::proforma, pappso::cbor::psm::PsmProtein::protein_sp, and pappso::cbor::psm::SageTsvHandler::Scan::psm_list.

Referenced by endLine().

◆ setCell()

void pappso::cbor::psm::SageTsvHandler::setCell ( const OdsCell & cell)
overridevirtual

callback that report the content of the current cell in a dedicated Cell object. Override it if you need to retrieve cell content.

Definition at line 87 of file sagetsvhandler.cpp.

88{
89 if(m_lineNumber == 0)
90 {
91 // header
92 QString tag = cell.getStringValue();
93 if(tag == "psm_id")
94 {
95 m_columnTypeList.push_back(Columns::psm_id);
96 }
97 else if(tag == "peptide")
98 {
99 m_columnTypeList.push_back(Columns::peptide);
100 }
101 else if(tag == "proteins")
102 {
103 m_columnTypeList.push_back(Columns::proteins);
104 }
105 else if(tag == "num_proteins")
106 {
107 m_columnTypeList.push_back(Columns::num_proteins);
108 }
109 else if(tag == "filename")
110 {
111 m_columnTypeList.push_back(Columns::filename);
112 }
113 else if(tag == "scannr")
114 {
115 m_columnTypeList.push_back(Columns::scannr);
116 }
117 else if(tag == "rank")
118 {
119 m_columnTypeList.push_back(Columns::rank);
120 }
121 else if(tag == "label")
122 {
123 m_columnTypeList.push_back(Columns::label);
124 }
125 else if(tag == "expmass")
126 {
127 m_columnTypeList.push_back(Columns::expmass);
128 }
129
130 else if(tag == "calcmass")
131 {
132 m_columnTypeList.push_back(Columns::calcmass);
133 }
134 else if(tag == "charge")
135 {
136 m_columnTypeList.push_back(Columns::charge);
137 }
138 else if(tag == "peptide_len")
139 {
140 m_columnTypeList.push_back(Columns::peptide_len);
141 }
142 else if(tag == "missed_cleavages")
143 {
144 m_columnTypeList.push_back(Columns::missed_cleavages);
145 }
146 else if(tag == "semi_enzymatic")
147 {
148 m_columnTypeList.push_back(Columns::semi_enzymatic);
149 }
150 else if(tag == "isotope_error")
151 {
152 m_columnTypeList.push_back(Columns::isotope_error);
153 }
154 else if(tag == "precursor_ppm")
155 {
156 m_columnTypeList.push_back(Columns::precursor_ppm);
157 }
158 else if(tag == "fragment_ppm")
159 {
160 m_columnTypeList.push_back(Columns::fragment_ppm);
161 }
162 else if(tag == "hyperscore")
163 {
164 m_columnTypeList.push_back(Columns::hyperscore);
165 }
166 else if(tag == "delta_next")
167 {
168 m_columnTypeList.push_back(Columns::delta_next);
169 }
170 else if(tag == "delta_best")
171 {
172 m_columnTypeList.push_back(Columns::delta_best);
173 }
174 else if(tag == "rt")
175 {
176 m_columnTypeList.push_back(Columns::rt);
177 }
178 else if(tag == "aligned_rt")
179 {
180 m_columnTypeList.push_back(Columns::aligned_rt);
181 }
182 else if(tag == "predicted_rt")
183 {
184 m_columnTypeList.push_back(Columns::predicted_rt);
185 }
186 else if(tag == "delta_rt_model")
187 {
188 m_columnTypeList.push_back(Columns::delta_rt_model);
189 }
190 else if(tag == "ion_mobility")
191 {
192 m_columnTypeList.push_back(Columns::ion_mobility);
193 }
194 else if(tag == "predicted_mobility")
195 {
196 m_columnTypeList.push_back(Columns::predicted_mobility);
197 }
198 else if(tag == "delta_mobility")
199 {
200 m_columnTypeList.push_back(Columns::delta_mobility);
201 }
202 else if(tag == "matched_peaks")
203 {
204 m_columnTypeList.push_back(Columns::matched_peaks);
205 }
206 else if(tag == "longest_b")
207 {
208 m_columnTypeList.push_back(Columns::longest_b);
209 }
210 else if(tag == "longest_y")
211 {
212 m_columnTypeList.push_back(Columns::longest_y);
213 }
214 else if(tag == "longest_y_pct")
215 {
216 m_columnTypeList.push_back(Columns::longest_y_pct);
217 }
218 else if(tag == "matched_intensity_pct")
219 {
220 m_columnTypeList.push_back(Columns::matched_intensity_pct);
221 }
222 else if(tag == "scored_candidates")
223 {
224 m_columnTypeList.push_back(Columns::scored_candidates);
225 }
226 else if(tag == "poisson")
227 {
228 m_columnTypeList.push_back(Columns::poisson);
229 }
230 else if(tag == "sage_discriminant_score")
231 {
232 m_columnTypeList.push_back(Columns::sage_discriminant_score);
233 }
234 else if(tag == "posterior_error")
235 {
236 m_columnTypeList.push_back(Columns::posterior_error);
237 }
238 else if(tag == "spectrum_q")
239 {
240 m_columnTypeList.push_back(Columns::spectrum_q);
241 }
242 else if(tag == "peptide_q")
243 {
244 m_columnTypeList.push_back(Columns::peptide_q);
245 }
246 else if(tag == "protein_q")
247 {
248 m_columnTypeList.push_back(Columns::protein_q);
249 }
250 else if(tag == "ms2_intensity")
251 {
252 m_columnTypeList.push_back(Columns::ms2_intensity);
253 }
254 else
255 {
256 throw pappso::ExceptionNotPossible(QObject::tr("column \"%1\" not defined").arg(tag));
257 }
258 }
259 else
260 {
261 if(m_columnNumber >= m_columnTypeList.size())
262 {
263 throw pappso::ExceptionOutOfRange(
264 QObject::tr("the value %1 is out of range").arg(cell.getStringValue()));
265 }
266 Columns column_type = m_columnTypeList[m_columnNumber];
267 switch(column_type)
268 {
269 case Columns::psm_id:
270 break;
271 case Columns::peptide:
272 parsePeptide(cell.toString());
273 break;
274 case Columns::proteins:
275 parseProteins(cell.toString());
276 break;
277 case Columns::num_proteins:
278 if((std::size_t)cell.getDoubleValue() != (std::size_t)m_proteinList.size())
279 {
280 throw pappso::PappsoException(
281 QObject::tr("column \"num_proteins\"!=%1").arg(m_proteinList.size()));
282 }
283 break;
284 case Columns::filename:
285 parseMsRunFilename(cell.toString());
286 break;
287 case Columns::scannr:
288 parseSpectrumStringId(cell.toString());
289 break;
290 case Columns::rank:
291 m_line.rank = cell.getDoubleValue();
292 break;
293 case Columns::label:
294 m_line.label = cell.getDoubleValue();
295 break;
296 case Columns::expmass:
297 m_line.expmass = cell.getDoubleValue();
298 break;
299 case Columns::calcmass:
300 m_line.calcmass = cell.getDoubleValue();
301 break;
302 case Columns::charge:
303 m_line.charge = cell.getDoubleValue();
304 break;
305 case Columns::peptide_len:
306 m_line.peptide_len = cell.getDoubleValue();
307 break;
308 case Columns::missed_cleavages:
309 m_line.missed_cleavages = cell.getDoubleValue();
310 break;
311 case Columns::semi_enzymatic:
312 m_line.semi_enzymatic = cell.getDoubleValue();
313 break;
314 case Columns::isotope_error:
315 m_line.isotope_error = cell.getDoubleValue();
316 break;
317
318 case Columns::precursor_ppm:
319 m_line.precursor_ppm = cell.getDoubleValue();
320 break;
321 case Columns::fragment_ppm:
322 m_line.fragment_ppm = cell.getDoubleValue();
323 break;
324 case Columns::hyperscore:
325 m_line.hyperscore = cell.getDoubleValue();
326 break;
327 case Columns::delta_next:
328 m_line.delta_next = cell.getDoubleValue();
329 break;
330 case Columns::delta_best:
331 m_line.delta_best = cell.getDoubleValue();
332 break;
333 case Columns::rt:
334 m_line.rt = cell.getDoubleValue() * 60; // to convert retention time in seconds
335 break;
336 case Columns::aligned_rt:
337 m_line.aligned_rt = cell.getDoubleValue();
338 break;
339 case Columns::predicted_rt:
340 m_line.predicted_rt = cell.getDoubleValue();
341 break;
342 case Columns::delta_rt_model:
343 m_line.delta_rt_model = cell.getDoubleValue();
344 break;
345 case Columns::ion_mobility:
346 m_line.ion_mobility = cell.getDoubleValue();
347 break;
348 case Columns::predicted_mobility:
349 m_line.predicted_mobility = cell.getDoubleValue();
350 break;
351 case Columns::delta_mobility:
352 m_line.delta_mobility = cell.getDoubleValue();
353 break;
354 case Columns::matched_peaks:
355 m_line.matched_peaks = cell.getDoubleValue();
356 break;
357 case Columns::longest_b:
358 m_line.longest_b = cell.getDoubleValue();
359 break;
360 case Columns::longest_y:
361 m_line.longest_y = cell.getDoubleValue();
362 break;
363 case Columns::longest_y_pct:
364 m_line.longest_y_pct = cell.getDoubleValue();
365 break;
366 case Columns::matched_intensity_pct:
367 m_line.matched_intensity_pct = cell.getDoubleValue();
368 break;
369 case Columns::scored_candidates:
370 m_line.scored_candidates = cell.getDoubleValue();
371 break;
372 case Columns::poisson:
373 m_line.poisson = cell.getDoubleValue();
374 break;
375 case Columns::sage_discriminant_score:
376 m_line.sage_discriminant_score = cell.getDoubleValue();
377 break;
378 case Columns::posterior_error:
379 m_line.posterior_error = cell.getDoubleValue();
380 break;
381 case Columns::spectrum_q:
382 m_line.spectrum_q = cell.getDoubleValue();
383 break;
384 case Columns::peptide_q:
385 m_line.peptide_q = cell.getDoubleValue();
386 break;
387 case Columns::protein_q:
388 m_line.protein_q = cell.getDoubleValue();
389 break;
390 case Columns::ms2_intensity:
391 m_line.ms2_intensity = cell.getDoubleValue();
392 break;
393 default:
394 qDebug() << "m_line.calcmass=" << m_line.calcmass;
395 throw pappso::ExceptionNotImplemented(
396 QObject::tr("column type %1 not implemented").arg((std::uint8_t)column_type));
397 break;
398 }
399
400 /*
401
402 2333 TMISDSDYTEFENFTK
403 GRMZM2G018197_P01;GRMZM2G068952_P01;GRMZM5G822976_P01 3
404 20120906_balliau_extract_1_A01_urnb-1.mzML controllerType=0
405 controllerNumber=1 scan=12542 1 1 1926.8225 1926.8193 2 16 0 0
406 0.0 1.6471838 1.9796097 54.06803492297634 28.049970383419556 0.0 38.192993
407 0.76385987 0.7671368 0.0032769442 0.0 0.0 0.0 16 2 14 0.875 32.54396 380
408 -13.375352220427656 1.1570586 -34.13482 0.00016041065 0.00022231363
409 0.00040124074 1271951.1
410 */
411 }
412 m_columnNumber++;
413}
pappso::cbor::psm::SageTsvHandler::parseProteins
void parseProteins(const QString &proteins_str)
Definition sagetsvhandler.cpp:475
pappso::cbor::psm::SageTsvHandler::parsePeptide
void parsePeptide(const QString &peptide_str)
Definition sagetsvhandler.cpp:438
pappso::cbor::psm::SageTsvHandler::parseSpectrumStringId
bool parseSpectrumStringId(const QString &spectrum_string_id)
Definition sagetsvhandler.cpp:498
pappso::cbor::psm::SageTsvHandler::parseMsRunFilename
void parseMsRunFilename(const QString &msrun_filename)
Definition sagetsvhandler.cpp:541
pappso::cbor::psm::SageTsvHandler::m_columnTypeList
std::vector< Columns > m_columnTypeList
Definition sagetsvhandler.h:222

References aligned_rt, calcmass, charge, delta_best, delta_mobility, delta_next, delta_rt_model, expmass, filename, fragment_ppm, hyperscore, ion_mobility, isotope_error, label, longest_b, longest_y, longest_y_pct, m_columnNumber, m_columnTypeList, m_line, m_lineNumber, m_proteinList, matched_intensity_pct, matched_peaks, missed_cleavages, ms2_intensity, num_proteins, parseMsRunFilename(), parsePeptide(), parseProteins(), parseSpectrumStringId(), peptide, peptide_len, peptide_q, poisson, posterior_error, precursor_ppm, predicted_mobility, predicted_rt, protein_q, proteins, psm_id, rank, rt, sage_discriminant_score, scannr, scored_candidates, semi_enzymatic, and spectrum_q.

◆ startLine()

void pappso::cbor::psm::SageTsvHandler::startLine ( )
overridevirtual

callback that indicates a new line start. Override it if needed.

Definition at line 416 of file sagetsvhandler.cpp.

417{
418 m_columnNumber = 0;
419 msp_peptide = nullptr;
420 m_proteinList.clear();
421 m_line = Line();
422}

References m_columnNumber, m_line, m_proteinList, and msp_peptide.

◆ startSheet()

void pappso::cbor::psm::SageTsvHandler::startSheet ( const QString & sheet_name)
overridevirtual

callback that indicates the begining of a data sheet. Override it in order to retrieve information about the current data sheet.

Definition at line 425 of file sagetsvhandler.cpp.

426{
427 m_columnNumber = 0;
428 m_lineNumber = 0;
429 mp_monitor->setStatus(QObject::tr("reading Sage TSV file"));
430
431 if(mp_monitor->shouldIstop())
432 {
433 throw pappso::ExceptionInterrupted(QObject::tr("Sage TSV data reading process interrupted"));
434 }
435}

References m_columnNumber, m_lineNumber, and mp_monitor.

◆ writePsm()

void pappso::cbor::psm::SageTsvHandler::writePsm ( const Psm & one_psm)
private

Definition at line 773 of file sagetsvhandler.cpp.

774{
775 m_sageReader.getCborStreamWriter().startMap();
776 m_sageReader.getCborStreamWriter().append("proforma");
777 m_sageReader.getCborStreamWriter().append(one_psm.proforma);
778 m_sageReader.getCborStreamWriter().append("protein_list");
779 QCborValue(one_psm.cbor_protein_list).toCbor(m_sageReader.getCborStreamWriter());
780
781 m_sageReader.getCborStreamWriter().append("eval");
782 m_sageReader.getCborStreamWriter().startMap();
783 m_sageReader.getCborStreamWriter().append("sage");
784 QCborValue(one_psm.cbor_eval).toCbor(m_sageReader.getCborStreamWriter());
785 m_sageReader.getCborStreamWriter().endMap();
786
787 m_sageReader.getCborStreamWriter().endMap();
788}

References pappso::cbor::psm::SageTsvHandler::Psm::cbor_eval, pappso::cbor::psm::SageTsvHandler::Psm::cbor_protein_list, m_sageReader, and pappso::cbor::psm::SageTsvHandler::Psm::proforma.

Referenced by writeScan().

◆ writeSample()

void pappso::cbor::psm::SageTsvHandler::writeSample ( const Sample & one_sample)
private

Definition at line 723 of file sagetsvhandler.cpp.

724{
725 m_sageReader.getCborStreamWriter().startMap();
726 m_sageReader.getCborStreamWriter().append("name");
727 one_sample.cbor_core_sample.value("name").toCbor(m_sageReader.getCborStreamWriter());
728
729 m_sageReader.getCborStreamWriter().append("identification_file_list");
730 one_sample.cbor_core_sample.value("identification_file_list")
731 .toCbor(m_sageReader.getCborStreamWriter());
732
733
734 m_sageReader.getCborStreamWriter().append("peaklist_file");
735 one_sample.cbor_core_sample.value("peaklist_file").toCbor(m_sageReader.getCborStreamWriter());
736 //"scan_list": [
737
738 m_sageReader.getCborStreamWriter().append("scan_list");
739 m_sageReader.getCborStreamWriter().startArray(one_sample.scan_map.size());
740 for(auto &it_scan : one_sample.scan_map)
741 {
742 writeScan(it_scan.second);
743 }
744 m_sageReader.getCborStreamWriter().endArray();
745
746
747 m_sageReader.getCborStreamWriter().endMap();
748}
pappso::cbor::psm::SageTsvHandler::writeScan
void writeScan(const Scan &one_scan)
Definition sagetsvhandler.cpp:751

References pappso::cbor::psm::SageTsvHandler::Sample::cbor_core_sample, m_sageReader, pappso::cbor::psm::SageTsvHandler::Sample::scan_map, and writeScan().

Referenced by endDocument().

◆ writeScan()

void pappso::cbor::psm::SageTsvHandler::writeScan ( const Scan & one_scan)
private

Definition at line 751 of file sagetsvhandler.cpp.

752{
753 m_sageReader.getCborStreamWriter().startMap();
754 m_sageReader.getCborStreamWriter().append("id");
755 QCborValue(one_scan.cbor_id).toCbor(m_sageReader.getCborStreamWriter());
756 m_sageReader.getCborStreamWriter().append("precursor");
757 QCborValue(one_scan.cbor_precursor).toCbor(m_sageReader.getCborStreamWriter());
758 m_sageReader.getCborStreamWriter().append("ms2");
759 QCborValue(one_scan.cbor_ms2).toCbor(m_sageReader.getCborStreamWriter());
760
761 m_sageReader.getCborStreamWriter().append("psm_list");
762 m_sageReader.getCborStreamWriter().startArray(one_scan.psm_list.size());
763 for(auto &it_psm : one_scan.psm_list)
764 {
765 writePsm(it_psm);
766 }
767 m_sageReader.getCborStreamWriter().endArray();
768
769 m_sageReader.getCborStreamWriter().endMap();
770}
pappso::cbor::psm::SageTsvHandler::writePsm
void writePsm(const Psm &one_psm)
Definition sagetsvhandler.cpp:773

References pappso::cbor::psm::SageTsvHandler::Scan::cbor_id, pappso::cbor::psm::SageTsvHandler::Scan::cbor_ms2, pappso::cbor::psm::SageTsvHandler::Scan::cbor_precursor, m_sageReader, pappso::cbor::psm::SageTsvHandler::Scan::psm_list, and writePsm().

Referenced by writeSample().

Member Data Documentation

◆ m_columnNumber

std::size_t pappso::cbor::psm::SageTsvHandler::m_columnNumber = 0
private

Definition at line 221 of file sagetsvhandler.h.

Referenced by setCell(), startLine(), and startSheet().

◆ m_columnTypeList

std::vector<Columns> pappso::cbor::psm::SageTsvHandler::m_columnTypeList
private

Definition at line 222 of file sagetsvhandler.h.

Referenced by setCell().

◆ m_decoyTag

QString pappso::cbor::psm::SageTsvHandler::m_decoyTag
private

Definition at line 229 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), and parseProteins().

◆ m_isFirstPass

bool pappso::cbor::psm::SageTsvHandler::m_isFirstPass
private

Definition at line 230 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), parseMsRunFilename(), parsePeptide(), parseProteins(), parseSpectrumStringId(), and recordLine().

◆ m_line

Line pappso::cbor::psm::SageTsvHandler::m_line
private

Definition at line 228 of file sagetsvhandler.h.

Referenced by endLine(), recordLine(), setCell(), and startLine().

◆ m_lineNumber

std::size_t pappso::cbor::psm::SageTsvHandler::m_lineNumber = 0
private

Definition at line 220 of file sagetsvhandler.h.

Referenced by endLine(), recordLine(), setCell(), and startSheet().

◆ m_progressIndex

std::size_t pappso::cbor::psm::SageTsvHandler::m_progressIndex = 0
private

Definition at line 219 of file sagetsvhandler.h.

Referenced by recordLine().

◆ m_proteinList

QStringList pappso::cbor::psm::SageTsvHandler::m_proteinList
private

Definition at line 231 of file sagetsvhandler.h.

Referenced by parseProteins(), recordLine(), setCell(), and startLine().

◆ m_psmProteinMap

PsmProteinMap& pappso::cbor::psm::SageTsvHandler::m_psmProteinMap
private

Definition at line 217 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), parseProteins(), and recordLine().

◆ m_sageReader

const SageReader& pappso::cbor::psm::SageTsvHandler::m_sageReader
private

Definition at line 216 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), parseMsRunFilename(), writePsm(), writeSample(), and writeScan().

◆ m_sampleMap

std::map<QString, Sample> pappso::cbor::psm::SageTsvHandler::m_sampleMap
private

Definition at line 233 of file sagetsvhandler.h.

Referenced by endDocument(), and parseMsRunFilename().

◆ m_spectrumIndex

std::size_t pappso::cbor::psm::SageTsvHandler::m_spectrumIndex
private

Definition at line 227 of file sagetsvhandler.h.

Referenced by parseSpectrumStringId(), and recordLine().

◆ m_spectrumNativeId

QString pappso::cbor::psm::SageTsvHandler::m_spectrumNativeId
private

Definition at line 232 of file sagetsvhandler.h.

Referenced by parseSpectrumStringId(), and recordLine().

◆ m_staticModificationList

std::vector<SageReader::SageModification> pappso::cbor::psm::SageTsvHandler::m_staticModificationList
private

Definition at line 225 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), and parsePeptide().

◆ m_variableModificationList

std::vector<SageReader::SageModification> pappso::cbor::psm::SageTsvHandler::m_variableModificationList
private

Definition at line 226 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), and parsePeptide().

◆ mp_currentSample

Sample* pappso::cbor::psm::SageTsvHandler::mp_currentSample
private

Definition at line 234 of file sagetsvhandler.h.

Referenced by parseMsRunFilename(), and recordLine().

◆ mp_monitor

pappso::UiMonitorInterface* pappso::cbor::psm::SageTsvHandler::mp_monitor = nullptr
private

Definition at line 218 of file sagetsvhandler.h.

Referenced by SageTsvHandler(), recordLine(), and startSheet().

◆ msp_peptide

pappso::PeptideSp pappso::cbor::psm::SageTsvHandler::msp_peptide
private

Definition at line 224 of file sagetsvhandler.h.

Referenced by parsePeptide(), recordLine(), and startLine().

The documentation for this class was generated from the following files: