The Feature Mapper node in the consensus workflow performs retention-time alignment and feature linking across data sets. It first chromatographically aligns the input files in a sample set. Specifically, it uses the results from the Minora algorithm to find the most abundant features and then aligns them to have the same retention time across all sample files. It uses the file in which most of these features are found as the reference for the rest of the alignment. For each file, the node tries to find a matching feature for every feature from the reference file. For each of these feature matches, it records the retention time difference in the two files. Finally, it fits a regression curve into the (RT, DRT) points. The node uses this curve to correct the retention times of the file relative to the chosen reference file.
For multidimensional data, the Feature Mapper node only aligns files having the same fraction number. It uses the grouping information in the Grouping & Quantification page in the study to align them. Before you start an analysis that uses the Feature Mapper node in the consensus workflow for label-free quantification, you can use this page to define sample groups of files that should be grouped together. Usually, you group files having equal properties, such as replicate files. See Grouping similar files for label free.
The Feature Mapper node subsequently links peptides and features and maps features across multiple files. It converts the LC/MS features from each individual raw data file into consensus features by matching retention time, monoisotopic m/z, and charge. It compares isotope patterns, retention time, and mass to determine if the peptide is the same. The node also tries to fill the gaps for files where it did not find a feature in the processing step. Only one data file needs an identified PSM for the Feature Mapper node to quantify a peptide across all files.
Whenever a feature is missing in one file but found in others—for example, in the case of features measured near the detection limit—the node’s feature-mapping algorithm identifies a feature gap and tries to fill it with an unused neighboring single peak. Each single peak used for gap-filling that is above the threshold specified by the node’s Single Peak Score Threshold parameter is included as a single-peak feature with the same charge state as all other features of this consensus feature.
The following figure shows the feature mapping on the left and its resulting consensus feature on the right, where:
- F stands for a feature.
- P stands for a single peak above the threshold.
- PT stands for a single peak below the threshold.
A consensus feature can consist of only single-peak features if there are no missing values or there is at least one PSM assigned to it. In a final step, the node assigns all features that cannot be identified in the processing step to consensus features. This step uses the parts per million and retention time tolerances set in the Feature Mapper node for grouping. The features must also have the same charge state.
The following figure shows the feature mapping on the left and its resulting consensus feature on the right, where P stands for a single peak above the threshold.
