generateCompounds: Automatic compound annotation

In rickhelmus/patRoon: Workflows for Mass-Spectrometry Based Non-Target Analysis

Automatic compound annotation

Description

Automatically perform chemical compound annotation for feature groups.

Usage

generateCompounds(fGroups, MSPeakLists, algorithm, ...)

## S4 method for signature 'featureGroups'
generateCompounds(fGroups, MSPeakLists, algorithm, ...)

Arguments

fGroups	featureGroups object which should be annotated. This should be the same or a subset of the object that was used to create the specified MSPeakLists. In the case of a subset only the remaining feature groups in the subset are considered.
MSPeakLists	A MSPeakLists object that was generated for the supplied fGroups.
algorithm	A character string describing the algorithm that should be used: "metfrag", "sirius", "library"
...	Any parameters to be passed to the selected compound generation algorithm.

Details

Several algorithms are provided to automatically perform compound annotation for feature groups. To this end, measured masses for all feature groups are searched within online database(s) (e.g. PubChem) to retrieve a list of potential candidate chemical compounds. Depending on the algorithm and its parameters, further scoring of candidates is then performed using, for instance, matching of measured and theoretical isotopic patterns, presence within other data sources such as patent databases and similarity of measured and in-silico predicted MS/MS fragments. Note that this process is often quite time consuming, especially for large feature group sets. Therefore, this is often one of the last steps within the workflow and not performed before feature groups have been prioritized.

generateCompounds is a generic function that will generateCompounds by one of the supported algorithms. The actual functionality is provided by algorithm specific functions such as generateCompoundsMetFrag and generateCompoundsSIRIUS. While these functions may be called directly, generateCompounds provides a generic interface and is therefore usually preferred.

Value

A compounds derived object containing all compound annotations.

Scorings

Each algorithm implements their own scoring system. Their names have been simplified and harmonized where possible. The compoundScorings function can be used to get an overview of both the algorithm specific and generic scoring names.

Sets workflows

With a sets workflow, annotation is first performed for each set. This is important, since the annotation algorithms typically cannot work with data from mixed ionization modes. The annotation results are then combined to generate a sets consensus:

• The annotation tables for each feature group from the set specific data are combined. Rows with overlapping candidates (determined by the first-block InChIKey) are merged.

• Set specific data (e.g. the ionic formula) is retained by renaming their columns with set specific names.

• The MS/MS fragment annotations (fragInfo column) from each set are combined.

• The scorings for each set are averaged to calculate overall scores. if setAvgSpecificScores=FALSE then scorings that are considered set specific (e.g. MS/MS and isotopic pattern match) are not averaged.

• The candidates are re-ranked based on their average ranking among the set data (if a candidate is absent in a set it is assigned the poorest rank in that set).

• The coverage of each candidate among sets is calculated. Depending on the setThreshold and setThresholdAnn arguments, candidates with low abundance are removed.

See Also

The compounds output class and its methods and the algorithm specific functions: generateCompoundsMetFrag, generateCompoundsSIRIUS, generateCompoundsLibrary

rickhelmus/patRoon documentation built on Nov. 22, 2024, 3:11 p.m.