R/generics.R, R/components-openms.R
generateComponentsOpenMS.RdUses the MetaboliteAdductDecharger utility (see http://www.openms.de) to generate components.
generateComponentsOpenMS(fGroups, ...)
# S4 method for class 'featureGroups'
generateComponentsOpenMS(
fGroups,
ionization = NULL,
chargeMin = 1,
chargeMax = 1,
chargeSpan = 3,
qTry = "heuristic",
potentialAdducts = NULL,
minRTOverlap = 0.66,
retWindow = 1,
absMzDev = 0.005,
minSize = 2,
relMinAdductAbundance = 0.75,
adductConflictsUsePref = TRUE,
NMConflicts = c("preferential", "mostAbundant", "mostIntense"),
prefAdducts = c("[M+H]+", "[M-H]-"),
extraOpts = NULL
)
# S4 method for class 'featureGroupsSet'
generateComponentsOpenMS(
fGroups,
ionization = NULL,
chargeMin = 1,
chargeMax = 1,
chargeSpan = 3,
qTry = "heuristic",
potentialAdducts = NULL,
...
)featureGroups object for which components should be generated.
Further arguments passed to the non-sets workflow method.
Which ionization polarity was used to generate the data: should be "positive"
or "negative". If the featureGroups object has adduct annotations, and ionization=NULL, the
ionization will be detected automatically.
This parameter is not supported for sets workflows, as the ionization will always be detected automatically.
The minimum/maximum charge to consider. Corresponds to the
algorithm:MetaboliteFeatureDeconvolution:charge_min/algorithm:MetaboliteFeatureDeconvolution:charge_min
options.
The maximum charge span for a single analyte. Corresponds to
algorithm:MetaboliteFeatureDeconvolution:charge_span_max.
Sets how charges are determined. Corresponds to algorithm:MetaboliteFeatureDeconvolution:q_try.
Valid options are "heuristic" and "all" (the "feature" option from OpenMS is
currently not supported).
The adducts to consider. Should be a numeric vector with probabilities for each
adduct, e.g. potentialAdducts=c("[M+H]+" = 0.8, "[M+Na]+" = 0.2). Note that the sum of probabilities
should always be 1. Furthermore, note that additions of multiple adducts should be controlled by the
chargeMin/chargeMax arguments (and not with potentialAdducts), e.g. if
chargeMax=2 then both [M+H]+ and [2M+H]2+ may be considered. Please see the
algorithm:MetaboliteFeatureDeconvolution:potential_adducts option of
MetaboliteAdductDecharger
for more details. If NULL then the a default is chosen with defaultOpenMSAdducts (which is
not the same as OpenMS).
Should be a list where each entry specifies the potential adducts for a set. Should either be named
with the sets names or follow the same order as sets(fGroups). Example:
potentialAdducts=list(positive=c("[M+H]+" = 0.8, "[M+Na]+" = 0.2), negative=c("[M-H]-" = 0.8, "[M-H2O-H]-" =
0.2))
Sets feature retention tolerances when grouping features. Sets the
"algorithm:MetaboliteFeatureDeconvolution:retention_max_diff" and
algorithm:MetaboliteFeatureDeconvolution:min_rt_overlap options.
Maximum absolute m/z deviation. Sets the algorithm:MetaboliteFeatureDeconvolution:mass_max_diff option
The minimum size of a component. Smaller components than this size will be removed. See note below.
The minimum relative abundance (0-1) that an adduct should be assigned to
features within the same feature group. See the Feature components section for more details.
If set to TRUE, and not all adduct assigments to the features within a feature
group are equal and at least one of those adducts is a preferential adduct (prefAdducts argument), then only
the features with (the lowest ranked) preferential adduct are considered. In all other cases or when
adductConflictsUsePref=FALSE only features with the most frequently assigned adduct is considered. See the
Feature components section for more details.
The strategies to employ when not all neutral masses within a component are equal. Valid options
are: "preferential", "mostAbundant" and "mostIntense". Multiple strategies are possible, and
will be executed in the given order until one succeeds. See the Feature components section for more details.
A character vector with one or more preferential adducts. See the Feature
components section for more details.
Named character vector with extra command line parameters directly passed to
MetaboliteAdductDecharger. Set to NULL to ignore.
A componentsFeatures derived object.
This function uses OpenMS to generate components. This function is called when calling generateComponents with
algorithm="openms".
Features that show highly similar chromatographic elution profiles are grouped, and subsequently annotated with their adducts.
The returned components are based on so called feature components. Unlike other algorithms, components are first made on a feature level (per analysis), instead of for complete feature groups. In the final step the feature components are converted to 'regular' components by employing a consensus approach with the following steps:
If an adduct assigned to a feature only occurs as a minority compared to other adduct assigments within the
same feature group, it is considered as an outlier and removed accordingly (controlled by the
relMinAdductAbundance argument).
For features within a feature group, only keep their adduct assignment if it occurs as the most frequent or
is preferential (controlled by adductConflictsUsePref and prefAdducts arguments).
Components are made by combining the feature groups for which at least one of their features are jointly present in the same feature component.
Conflicts of neutral mass assignments within a component (i.e. not all are the same) are dealt with.
Firstly, all feature groups with an unknown neutral mass are split in another component. Then, if conflicts still
occur, the feature groups with similar neutral mass (determined by absMzDev argument) are grouped. Depending
on the NMConflicts argument, the group with one or more preferential adduct(s) or that is the largest or
most intense is selected, whereas others are removed from the component. In case multiple groups contain
preferential adducts, and >1 preferential adducts are available, the group with the adduct that matches
first in prefAdducts 'wins'. In case of ties, one of the next strategies in NMConflicts is tried.
If a feature group occurs in multiple components it will be removed completely.
the minSize filter is applied.
In a sets workflow the componentization is first performed for each
set independently. The resulting components are then all combined in a componentsSet object. Note that
the components themselves are never merged. The components are renamed to include the set name from which they were
generated (e.g. "CMP1" becomes "CMP1-positive").
generateComponentsOpenMS uses multiprocessing to parallelize computations. Please see the parallelization section in the handbook for more details and patRoon options for configuration options.
Bielow2010patRoon
generateComponents for more details and other algorithms.