Retreats Spring2014 OpenMS2.0

Protocol of the OpenMS 2.0 discussion

monday afternoon session

• while originally planned we will not be able to do the OpenMS 2.0 implementation in a single attempt (e.g., individual branch that is later on merged)
• we want to collect a list of features for OpenMS 2.0 during the retreat that can afterwards be implemented one by one
  • issues in the github tracker
• a crucial part will be benchmarking of the changes (do we improve anything beyond design, or does it get worse)
• what should we measure (cpu time, memory consumption)
• automatic vs. manually
• we will define a benchmark suite for specific algorithms and data structures that can be run easily
• We start to go through the individual KERNEL classes
  • AreaIterator = we will check it's usage but most likely remove it
• Peak (basically breaks the discussion on individual classes)
• we discuss if we should switch from class Peak to pairs of std::vector<double> representing m/z and intensity [everybody agrees]

  class ISpectrum 
  {
    std::vector<double> mz_;
    std::vector<double> int_;

    // + associative container for meta-data arrays (e.g., float-binary-data)
    map<MetaDataID, MetaDataArray> meta_;
  }

• additional data should be hold in SpectrumSettings objects and not directly in the spectrum
• Which meta information are essential and hence should be added directly to the spectrum (everybody will name his favourites)
  • check the mzMl standard what is mandatory for a spectrum
• What should we use as ID for our spectrum and how can we make sure that is correctly mapped to the mzML and stays correct?

thursday morning session

• intensive discussion on spectrum interface, whiteboard state
• intensive discussion on msexperiment interface

class ISpectraAccess 
{
public:
  int getNrSpectra() const;
  getMetaSpectrum(int i) const;
  const ISpectrum& getSpectrum(int i) const;
  const ExpMeta& getExperimentMeta() const;
  // iterators for spectra
  SpectrumIterator spectrumBegin();
  SpectrumIterator spectrumEnd();
};

class IChromatogramAccess 
{
public:
  int getNrChromatograms() const;
  getMetaSpectrum(int i) const;
  const IChromatogram& getChromatogram(int i) const;
  const ExpMeta& getExperimentMeta() const;
  // iterators for chromatograms
  ChromatogramIterator chromatogramBegin();
  ChromatogramIterator chromatogramEnd();
};

• MSExperiment will be renamed to MSRun, whiteboard state
• discussion on how to proceed with features/consensusfeatures
  • should we use the mzQuantML data model
    • too much information
  • how can we model relations between entities (e.g., RDF triple store)
  • how can we model relations between the individual experiments (or channels)
• we will remove consensusfeature
  • MassTrace
  • Feature
  • MRMFeature
  • Map whiteboard state
• MetaInformation discussion

wednesday morning session

• agreed on further modularisation of OpenMS
• support utils with more classes as an incubator for OpenMS/TOPP
• prepare small libOpenMSChemistry as a container for lightweight future developments whiteboard state