Skip to contents

Functionality to predict quantitative data

Source: R/generics.R, R/main.R, R/feature_groups.R, and 9 more
pred-quant.Rd

Functions to predict response factors and feature concentrations from SMILES and/or SIRIUS+CSI:FingerID fingerprints using the MS2Quant package.

Usage

calculateConcs(fGroups, ...)

# S4 method for class 'featureGroups'
calculateConcs(fGroups, featureAnn, areas = FALSE)

# S4 method for class 'featureGroupsSet'
calculateConcs(fGroups, featureAnn, areas = FALSE)

# S4 method for class 'compounds'
predictRespFactors(
  obj,
  fGroups,
  calibrants,
  eluent,
  organicModifier,
  pHAq,
  concUnit = "ugL",
  calibConcUnit = concUnit,
  updateScore = FALSE,
  scoreWeight = 1,
  parallel = TRUE
)

# S4 method for class 'featureGroupsScreening'
predictRespFactors(
  obj,
  calibrants,
  eluent,
  organicModifier,
  pHAq,
  concUnit = "ugL",
  calibConcUnit = concUnit
)

# S4 method for class 'featureGroupsScreening'
calculateConcs(fGroups, featureAnn = NULL, areas = FALSE)

# S4 method for class 'featureGroupsScreeningSet'
predictRespFactors(obj, calibrants, ...)

# S4 method for class 'featureGroupsScreeningSet'
calculateConcs(fGroups, featureAnn = NULL, areas = FALSE)

# S4 method for class 'compoundsSet'
predictRespFactors(obj, fGroups, calibrants, ...)

# S4 method for class 'compoundsSIRIUS'
predictRespFactors(
  obj,
  fGroups,
  calibrants,
  eluent,
  organicModifier,
  pHAq,
  concUnit = "ugL",
  calibConcUnit = concUnit,
  type = "FP"
)

# S4 method for class 'formulasSet'
predictRespFactors(obj, fGroups, calibrants, ...)

# S4 method for class 'formulasSIRIUS'
predictRespFactors(
  obj,
  fGroups,
  calibrants,
  eluent,
  organicModifier,
  pHAq,
  concUnit = "ugL",
  calibConcUnit = concUnit
)

getQuantCalibFromScreening(
  fGroups,
  concs,
  areas = FALSE,
  average = FALSE,
  IMS = "maybe"
)

Arguments

fGroups

For predictRespFactors methods for feature annotations: The featureGroups object for which the annotations were performed.

For calculateConcs: The featureGroups object for which concentrations should be calculated.

For getQuantCalibFromScreening: A feature groups object screened for the calibrants with screenSuspects.

...

(sets workflow) Further arguments passed to the non-sets workflow method.

featureAnn

A featureAnnotations object (e.g. formulasSIRIUS or compounds) which contains response factors. Optional if calculateConcs is called on suspect screening results (i.e. featureGroupsScreening method).

areas

Set to TRUE to use peak areas instead of peak heights. Note: for calculateConcs this should follow what is in the calibrants table.

obj

The workflow object for which predictions should be performed, e.g. feature groups with screening results (featureGroupsScreening) or compound annotations (compounds).

calibrants

A data.frame with calibrants, see the Calibration section below.

(sets workflow) Should be a list with the calibrants for each set.

eluent

A data.frame that describes the LC gradient program. Should have a column time with the retention time in seconds and a column B with the corresponding percentage of the organic modifier (0-100).

organicModifier

The organic modifier of the mobile phase: either "MeOH" (methanol) or "MeCN" (acetonitrile).

pHAq

The pH of the aqueous part of the mobile phase.

concUnit

The concentration unit for calculated concentrations. Can be molar based ("nM", "uM", "mM", "M") or mass based ("ngL", "ugL", "mgL", "gL"). Furthermore, can be prefixed with "log " for logarithmic concentrations (e.g. "log mM").

calibConcUnit

The concentration unit used in the calibrants table. For possible values see the concUnit argument.

updateScore, scoreWeight

If updateScore=TRUE then the annotation score column is updated by adding normalized values of the response factor (weighted by scoreWeight). Currently, this only makes sense for annotations performed with MetFrag!

parallel

If set to TRUE then code is executed in parallel through the future package. Please see the parallelization section in the handbook for more details.

type

Which types of predictions should be performed: should be "FP" (SIRIUS+CSI:FingerID fingerprints), "SMILES" or "both". Only relevant for compoundsSIRIUS method.

concs

A data.frame with concentration data. See the Calibration section below.

average

Set to TRUE to average intensity values within replicates.

IMS

(IMS workflow) Specifies which feature groups are considered for generating calibrant data in IMS workflows. The following options are valid:

  • "both": Selects IMS and non-IMS features.

  • "maybe": Selects non-IMS features and IMS features without assigned IMS precursor.

  • FALSE: Selects only non-IMS features.

  • TRUE: Selects only IMS features.

Best to keep this "maybe", as calibration typically doesn't support IMS filtered data.

Value

predictRespFactors returns an object amended with response factors (RF_SMILES/LRF_SIRFP columns).

calculateConcs returns a featureGroups based object amended with concentrations for each feature group (accessed with the concentrations method).

Details

The MS2Quant R package predicts concentrations from SMILES and/or MS/MS fingerprints obtained with SIRIUS+CSI:FingerID. The predictRespFactors method functions interface with this package to calculate response factors, which can then be used to calculate feature concentrations with the calculateConcs method function.

Note

The rcdk package and OpenBabel tool are used internally to calculate molecular weights. Please make sure that OpenBabel is installed.

MS2Quant currently only supports M+H and M+ adducts when performing predictions with SIRIUS:FingerID fingerprints. Predictions for candidates with other adducts, including M-H, are skipped with a warning.

Calibration

The MS2Quant package requires calibration to convert predicted ionization efficiencies to instrument/method specific response factors. The calibration data should be specified with the calibrants argument to predictRespFactors. This should be a data.frame with intensity observations at different concentrations for a set of calibrants. Each row specifies one intensity observation at one concentration. The table should have the following columns:

  • name The name of the calibrant. Can be freely chosen.

  • SMILES The SMILES of the calibrant.

  • rt The retention time of the calibrant (in seconds).

  • intensity The peak intensity (or area, see the areas argument) of the calibrant.

  • conc The concentration of the calibrant (see the calibConcUnit argument for specifying the unit).

It is recommended to include multiple calibrants (e.g. >=10) at multiple concentrations (e.g. >=5). The latter is achieved by adding multiple rows for the same calibrant (keeping the name/SMILES/rt columns constant). It is also possible to follow the column naming used by MS2Quant (however retention times should still be in seconds!). For more details and tips see https://github.com/kruvelab/MS2Quant.

The getQuantCalibFromScreening function can be used to automatically generate a calibrants table from a feature groups object with suspect screening results. Here, the idea is to perform a screening with screenSuspects with a suspect list that contain the calibrants, which is then used to construct the calibrant table. It is highly recommended to add retention times for the calibrants in the suspect list to ensure the calibrant is assigned to the correct feature. Furthermore, it is possible to simply add the calibrants to the 'regular' suspect list in case a suspect screening was already part of the workflow. The getQuantCalibFromScreening function still requires you to specify concentration data, which is achieved via the concs argument. This should be a data.frame with a column name corresponding to the calibrant name (i.e. same as used by screenSuspects above) and columns with concentration data. The latter columns specify the concentrations of a calibrant in different replicates (as defined in the analysis information). The concentration columns should be named after the corresponding replicate. Only those replicates that should be used for calibration need to be included. Furthermore, NA values can be used if a replicate should be ignored for a specific calibrant.

Predicting response factors

The response factors are predicted with the predictRespFactors generic functions, which accepts the following input:

  • Suspect screening results. The SMILES data is used to predict response factors for suspect hits.

  • Formula annotation data obtained with "sirius" algorithm (generateFormulasSIRIUS). The predictions are performed for each formula candidate using SIRIUS+CSI:FingerID fingerprints. For this reason, the getFingerprint argument must be set to TRUE when generating the formula data.

  • Compound annotation data obtained with the "sirius" algorithm (generateCompoundsSIRIUS). The predictions are performed for each annotation candidate using its SMILES and/or SIRIUS+CSI:FingerID fingerprints. The predictions are performed on a per formula basis, hence, response factors for isomers will be equal.

  • Compound annotation data obtained with algorithms other than "sirius". The response factors are predicted from SMILES data.

When SMILES data is used then predictions of response factors are generally more accurate. However, calculations with SIRIUS+CSI:FingerID fingerprints are faster and only require the formula and MS/MS spectrum, i.e. not the full structure. Hence, calculations with SMILES are mostly useful in suspect screening workflows, or with high confidence compound annotation data, whereas MS/MS fingerprints are suitable with unknowns.

For annotation data the calculations are performed for all candidates. This can especially lead to long running calculations when SMILES data is used. Hence, it is strongly recommended to first prioritize the annotation results, e.g. with the topMost argument to the filter method.

When response factors are predicted from SIRIUS+CSI:FingerID fingerprints then only formula and MS/MS spectra are used, even if compound annotations are used for input. The major difference is that with formula annotation input all formula candidates for which a fingerprint could be generated are considered, whereas with compound annotations only candidate formulae are considered for which also a structure could be assigned. Hence, the formula annotation input could be more comprehensive, whereas predictions from structure annotations could lead to more representative results as only formulae are considered for which at least one structure could be assigned.

Assigning concentrations

The calculateConcs generic function is used to assign concentrations for each feature using the response factors discussed in the previous section. The function takes response factors from suspect screening results and/or feature annotation data. If multiple response factors were predicted for the same feature group, for instance when multiple annotation candidates or suspect hits for this feature group are present, then a concentrations is assigned for all response factors. These values can later be easily aggregated with e.g. the as.data.table function.

In IMS workflows with post mobility assignments (see assignMobilities), the intensities of IMS precursors are used for the calculation of concentrations for IMS features (as calibration typically does not consider differences due to mobility filtering). However, the response factor assigned to IMS features are still used. This may yield small differences compared to workflows where concentrations are assigned prior to mobility assignments, as assignMobilities simply copies concentrations from IMS precursors to IMS features.

References

OBoyle NM, Banck M, James CA, Morley C, Vandermeersch T, Hutchison GR (2011). “Open Babel: An open chemical toolbox.” Journal of Cheminformatics, 3(1). doi:10.1186/1758-2946-3-33 .

Guha R (2007). “Chemical Informatics Functionality in R.” Journal of Statistical Software, 18(6).

Sepman H, Malm L, Peets P, MacLeod M, Martin J, Breitholtz M, Kruve A (2023). “Bypassing the Identification: MS2Quant for Concentration Estimations of Chemicals Detected with Nontarget LC-HRMS from MS2 Data.” Analytical Chemistry, 95(33), 12329–12338. doi:10.1021/acs.analchem.3c01744 . https://doi.org/10.1021/acs.analchem.3c01744.

See also

Toxicity prediction