utils.cpp

#include "utils.h"
#include <pappsomspp/exception/exceptionnotfound.h>
#include <pappsomspp/mass_range.h>
#include <cmath>

const QUrl
Utils::getOlsUrl(QString psimod_accession)
{

  QString iri(QString("http://purl.obolibrary.org/obo/%1")
                .arg(psimod_accession.replace(":", "_")));
  QUrl url(
    QString("http://www.ebi.ac.uk/ols/ontologies/mod/terms?iri=%1").arg(iri));
  return url;
}

const QString
Utils::getXmlDouble(pappso::pappso_double number)
{
  return QString::number(number, 'g', 6);
}

const QString
Utils::getIdentificationEngineName(IdentificationEngine engine)
{
  QString engine_name;
  switch(engine)
    {
      case IdentificationEngine::XTandem:
        engine_name = "X!Tandem";
        break;
      case IdentificationEngine::mascot:
        engine_name = "Mascot";
        break;

      case IdentificationEngine::peptider:
        engine_name = "Peptider";
        break;
      case IdentificationEngine::OMSSA:
        engine_name = "OMSSA";
        break;
      case IdentificationEngine::SEQUEST:
        engine_name = "Sequest";
        break;
      case IdentificationEngine::Comet:
        engine_name = "Comet";
        break;
      case IdentificationEngine::Morpheus:
        engine_name = "Morpheus";
        break;
      case IdentificationEngine::MSGFplus:
        engine_name = "MS-GF+";
        break;
    }
  return engine_name;
}

const QString
Utils::getDatabaseName(ExternalDatabase database)
{
  QString database_name;
  switch(database)
    {
      case ExternalDatabase::AGI_LocusCode:
        database_name = "AGI_LocusCode";
        break;
      case ExternalDatabase::NCBI_gi:
        database_name = "NCBI_gi";
        break;

      case ExternalDatabase::SwissProt:
        database_name = "Swiss-Prot";
        break;
      case ExternalDatabase::TrEMBL:
        database_name = "TrEMBL";
        break;
      case ExternalDatabase::ref:
        database_name = "ref";
        break;
    }
  return database_name;
}


std::vector<std::pair<pappso::pappso_double, size_t>>
Utils::getHistogram(std::vector<pappso::pappso_double> data_values,
                    unsigned int number_of_class)
{
  std::vector<std::pair<pappso::pappso_double, size_t>> histogram(
    number_of_class + 1);
  try
    {
      qDebug() << "Utils::getHistogram begin";

      std::sort(data_values.begin(), data_values.end());
      pappso::pappso_double min   = *data_values.begin();
      pappso::pappso_double max   = *(data_values.end() - 1);
      pappso::pappso_double total = std::abs(max - min);
      pappso::pappso_double offset =
        (total / (pappso::pappso_double)number_of_class);
      // qDebug() << "Utils::getHistogram number_of_class offset=" << offset;
      for(unsigned int i = 0; i < histogram.size(); i++)
        {
          histogram[i] = std::pair<pappso::pappso_double, size_t>{
            (min + (offset * i) + (offset / 2)), 0};
          // qDebug() << "Utils::getHistogram x=" << histogram[i].first;
        }
      // qDebug() << "Utils::getHistogram data_values";
      for(pappso::pappso_double value : data_values)
        {
          // qDebug() << "Utils::getHistogram value=" << value;
          unsigned int i = std::abs((value - min) / offset);
          // qDebug() << "Utils::getHistogram i=" << i << " size=" <<
          // histogram.size();
          histogram.at(i).second++;
        }
    }
  catch(std::exception exception_std)
    {
      throw pappso::PappsoException(
        QObject::tr("Utils::getHistogram error %1").arg(exception_std.what()));
    }
  qDebug() << "Utils::getHistogram end";
  return histogram;
}


pappso::AaModificationP
Utils::guessAaModificationPbyMonoisotopicMassDelta(pappso::mz mass)
{
  pappso::PrecisionP precision = pappso::Precision::getDaltonInstance(0.01);

  pappso::AaModificationP oxidation =
    pappso::AaModification::getInstance("MOD:00719");
  if(pappso::MassRange(oxidation->getMass(), precision).contains(mass))
    {
      return oxidation;
    }
  pappso::AaModificationP iodoacetamide =
    pappso::AaModification::getInstance("MOD:00397");
  if(pappso::MassRange(iodoacetamide->getMass(), precision).contains(mass))
    {
      return iodoacetamide;
    }
  pappso::AaModificationP acetylated =
    pappso::AaModification::getInstance("MOD:00408");
  if(pappso::MassRange(acetylated->getMass(), precision).contains(mass))
    {
      return acetylated;
    }
  pappso::AaModificationP phosphorylated =
    pappso::AaModification::getInstance("MOD:00696");
  if(pappso::MassRange(phosphorylated->getMass(), precision).contains(mass))
    {
      return phosphorylated;
    }
  pappso::AaModificationP ammonia =
    pappso::AaModification::getInstance("MOD:01160");
  if(pappso::MassRange(ammonia->getMass(), precision).contains(mass))
    {
      return ammonia;
    }
  pappso::AaModificationP dehydrated =
    pappso::AaModification::getInstance("MOD:00704");
  if(pappso::MassRange(dehydrated->getMass(), precision).contains(mass))
    {
      return dehydrated;
    }
  pappso::AaModificationP dimethylated =
    pappso::AaModification::getInstance("MOD:00429");
  if(pappso::MassRange(dimethylated->getMass(), precision).contains(mass))
    {
      return dimethylated;
    }

  pappso::AaModificationP dimethylated_medium =
    pappso::AaModification::getInstance("MOD:00552");
  if(pappso::MassRange(dimethylated_medium->getMass(), precision)
       .contains(mass))
    {
      return dimethylated_medium;
    }

  pappso::AaModificationP dimethylated_heavy =
    pappso::AaModification::getInstance("MOD:00638");
  if(pappso::MassRange(dimethylated_heavy->getMass(), precision).contains(mass))
    {
      return dimethylated_heavy;
    }
  pappso::AaModificationP DimethylpyrroleAdduct =
    pappso::AaModification::getInstance("MOD:00628");
  if(pappso::MassRange(DimethylpyrroleAdduct->getMass(), precision)
       .contains(mass))
    {
      return DimethylpyrroleAdduct;
    }

  // modification not found, creating customized mod :
  return pappso::AaModification::getInstanceCustomizedMod(mass);

  throw pappso::ExceptionNotFound(
    QObject::tr("Utils::guessAaModificationPbyMonoisotopicMassDelta => "
                "modification not found for mass %1")
      .arg(mass));
}


pappso::AaModificationP
Utils::translateAaModificationFromUnimod(const QString &unimod_accession)
{
  if(unimod_accession == "UNIMOD:1")
    {

      return pappso::AaModification::getInstance("MOD:00394");
    }
  if(unimod_accession == "UNIMOD:4")
    {

      return pappso::AaModification::getInstance("MOD:00397");
    }
  if(unimod_accession == "UNIMOD:27")
    {

      return pappso::AaModification::getInstance("MOD:00420");
    }
  // UNIMOD:28 => MOD:00040
  if(unimod_accession == "UNIMOD:28")
    {

      return pappso::AaModification::getInstance("MOD:00040");
    }

  if(unimod_accession == "UNIMOD:35")
    {

      return pappso::AaModification::getInstance("MOD:00425");
    }
  qDebug() << __FILE__ << " " << __FUNCTION__ << " " << __LINE__ << " "
           << unimod_accession << " not found";
  return nullptr;
}