Auto irt

src/openms/include/OpenMS/ANALYSIS/OPENSWATH/OpenSwathHelper.h

@@ -198,6 +198,22 @@ namespace OpenMS
     static std::map<std::string, double> simpleFindBestFeature(const OpenMS::MRMFeatureFinderScoring::TransitionGroupMapType & transition_group_map, 
                                                                bool useQualCutoff = false,
                                                                double qualCutoff = 0.0);
+
+    /**
+     * @brief Subsample a library based on a given 
+     * 
+     * This method is used to subsample the library intelligently in order to ensure that the subsampled compounds are distributed across RT space.
+     * 
+     * @param nrBins Number of bins across RT 
+     * @param peptidesPerBin aim number of peptides per bin, if not reached a warning is issued
+     *
+     * @return LightTargetedExperiment with the subsampled library
+     */
+    static OpenSwath::LightTargetedExperiment subsampleLibrary(OpenSwath::LightTargetedExperiment& library,
+      int nrBins,
+      int minPeptidesPerBin);
+
+
   };
 
 } // namespace OpenMS

src/openms/include/OpenMS/ANALYSIS/OPENSWATH/OpenSwathWorkflow.h

@@ -280,6 +280,20 @@ namespace OpenMS
       bool pasef = false,
       bool load_into_memory = false);
 
+    TransformationDescription performAutoRTNormalization(
+      OpenSwath::LightTargetedExperiment& irt_transitions,
+      std::vector< OpenSwath::SwathMap > & swath_maps,
+      TransformationDescription& im_trafo,
+      double min_rsq,
+      double min_coverage,
+      const Param& feature_finder_param,
+      ChromExtractParams& cp_irt,
+      const Param& irt_detection_param,
+      const Param& calibration_param,
+      Size debug_level,
+      bool pasef,
+      bool load_into_memory);
+
   public:
 
     /** @brief Perform retention time and m/z calibration

src/openms/source/ANALYSIS/OPENSWATH/OpenSwathHelper.cpp

@@ -7,6 +7,7 @@
 // --------------------------------------------------------------------------
 
 #include <OpenMS/ANALYSIS/OPENSWATH/OpenSwathHelper.h>
+#include <OpenMS/MATH/STATISTICS/Histogram.h>
 
 namespace OpenMS
 {
@@ -187,4 +188,68 @@ namespace OpenMS
     return result;
   }
 
+  OpenSwath::LightTargetedExperiment OpenSwathHelper::subsampleLibrary(OpenSwath::LightTargetedExperiment& library,
+                                                               int nrBins, 
+                                                               int peptidesPerBin)
+  {
+    std::pair<double,double> RTRange = OpenSwathHelper::estimateRTRange(library);
+    OPENMS_LOG_DEBUG << "Detected retention time range from " << RTRange.first << " to " << RTRange.second << std::endl;
+
+    // Sort the LightTargetedExperiment Peptide precursors by their RT
+    auto peptide_precursors = library.getCompounds();
+    std::sort(peptide_precursors.begin(), peptide_precursors.end(),
+              [](const OpenSwath::LightCompound& a, const OpenSwath::LightCompound& b)
+              {
+                return a.rt < b.rt;
+              });
+
+    std::vector<OpenSwath::LightCompound> selected_compounds;
+
+    // Create a histogram bins with the desired number of bins
+    Math::Histogram<> hist(RTRange.first, RTRange.second, nrBins);
+    for (size_t h = 0; h < hist.size(); h++)
+    {
+      OPENMS_LOG_DEBUG << "Bin " << h << " has left border of " << hist.leftBorderOfBin(h) << " and right border of " << hist.rightBorderOfBin(h) << std::endl;
+    }
+
+    // Take the first `minPeptidesPerBin` peptides from each bin
+    int i = 0; // index of the peptide precursor
+    for (size_t h = 0; h < hist.size(); h++)
+    {
+      int numPepsInBin = 0;
+      bool doneWithBin = false;
+      while ((numPepsInBin < peptidesPerBin) && (!doneWithBin)) 
+      {
+        // Check if the peptide precursor is within the bin range
+        OPENMS_LOG_DEBUG << "Checking peptide precursor with RT " << peptide_precursors[i].rt << " against bin with left border of " << hist.leftBorderOfBin(h) << std::endl;
+        if (peptide_precursors[i].rt >= hist.leftBorderOfBin(h) && peptide_precursors[i].rt < hist.rightBorderOfBin(h))
+        {
+          OPENMS_LOG_DEBUG << "Adding peptide precursor with RT " << peptide_precursors[i].rt << " to bin with left border of " << hist.leftBorderOfBin(h) << std::endl;
+          selected_compounds.push_back(peptide_precursors[i]);
+          numPepsInBin++;
+          i++;
+        }
+        else
+        {
+          // If we have less than the minimum number of peptides in the bin, warn the user
+          OPENMS_LOG_WARN << "RT Bin from " << hist.leftBorderOfBin(h) << " to " << hist.rightBorderOfBin(h) << " has only " 
+          << numPepsInBin << " peptides. This is less than the minumum number specified of " << peptidesPerBin << std::endl;
+
+          doneWithBin = true;
+        }
+      }
+
+      // Have the minumum number of peptides required, seek to the next bin
+      while (peptide_precursors[i].rt < hist.rightBorderOfBin(h))
+      {
+        i++;
+      }
+      numPepsInBin = 0; 
+    }
+
+    OPENMS_LOG_DEBUG << "Subsampled library has " << selected_compounds.size() << " precursors" << std::endl;
+    return library.selectCompounds(selected_compounds);
+  }
+
+
 }