MetFragR
MetFrag is available as R-package.
MetFragR on github Submit an issue
The R package enables functionalities from the MetFrag Commandline tool to be used within the R programming language.
Install
Local Install
First, check out the MetFragR GitHub repository and build the package (on command line):
| git clone https://github.com/ipb-halle/MetFragR.git |
| cd MetFragR |
| R CMD check metfRag |
| R CMD build metfRag |
After the succesful build turn into R and install the package:
| install.packages("metfRag",repos=NULL,type="source") |
| library(metfRag) |
Net Install (over GitHub)
The easiest way to install MetFragR is to use the GitHub link:
| library(devtools) |
| install_github("c-ruttkies/MetFragR/metfRag") |
| library(metfRag) |
Example
The following example shows how to run a simple MetFrag query in R.
| # |
| # first define the settings object |
| # |
| settingsObject<-list() |
| # |
| # set database parameters to select candidates |
| # |
| settingsObject[["DatabaseSearchRelativeMassDeviation"]]<-5.0 |
| settingsObject[["FragmentPeakMatchAbsoluteMassDeviation"]]<-0.001 |
| settingsObject[["FragmentPeakMatchRelativeMassDeviation"]]<-5.0 |
| settingsObject[["MetFragDatabaseType"]]<-"PubChem" |
| # |
| # the more information about the precurosr is available |
| # the more precise is the candidate selection |
| # |
| settingsObject[["NeutralPrecursorMass"]]<-253.966126 |
| settingsObject[["NeutralPrecursorMolecularFormula"]]<-"C7H5Cl2FN2O3" |
| settingsObject[["PrecursorCompoundIDs"]]<-c("50465", "57010914", "56974741", "88419651", "23354334") |
| # |
| # pre and post-processing filter |
| # |
| # define filters to filter unconnected compounds (e.g. salts) |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("UnconnectedCompoundFilter","IsotopeFilter") |
| settingsObject[["MetFragPostProcessingCandidateFilter"]]<-c("InChIKeyFilter") |
| # |
| # define the peaklist as 2-dimensional matrix |
| # |
| settingsObject[["PeakList"]]<-matrix(c( |
| 90.97445, 681, |
| 106.94476, 274, |
| 110.02750, 110, |
| 115.98965, 95, |
| 117.98540, 384, |
| 124.93547, 613, |
| 124.99015, 146, |
| 125.99793, 207, |
| 133.95592, 777, |
| 143.98846, 478, |
| 144.99625, 352, |
| 146.00410, 999, |
| 151.94641, 962, |
| 160.96668, 387, |
| 163.00682, 782, |
| 172.99055, 17, |
| 178.95724, 678, |
| 178.97725, 391, |
| 180.97293, 999, |
| 196.96778, 720, |
| 208.96780, 999, |
| 236.96245, 999, |
| 254.97312, 999), ncol=2, byrow=TRUE) |
| # |
| # run MetFrag |
| # |
| scored.candidates<-run.metfrag(settingsObject) |
| # |
| # scored.candidates is a data.frame with scores and candidate properties |
| # |
Candidate Filters
Pre-pocessing Candidate Filters
Filters can be defined to filter candidates prior to fragmentation. Following filters are available:
| UnconnectedCompoundFilter | - | filter non-connected compounds (e.g. salts) |
| IsotopeFilter | - | filter compounds containing non-standard isotopes |
| MinimumElementsFilter | - | filter by minimum of contained elements |
| MaximumElementsFilter | - | filter by maximum of contained elements |
| SmartsSubstructureInclusionFilter | - | filter by presence of defined sub-structures |
| SmartsSubstructureExclusionFilter | - | filter by absence of defined sub-structures |
| ElementInclusionFilter | - | filter by presence of defined elements (other elements are allowed) |
| ElementInclusionExclusiveFilter | - | filter by presence of defined elements (no other elements are allowed) |
| ElementExclusionFilter | - | filter by absence of defined sub-structures |
When defining pre-processing filters further parameters have to be defined:
| MinimumElementsFilter | - | FilterMinimumElements |
| MaximumElementsFilter | - | FilterMaximumElements |
| SmartsSubstructureInclusionFilter | - | FilterSmartsInclusionList |
| SmartsSubstructureExclusionFilter | - | FilterSmartsExclusionList |
| ElementInclusionFilter | - | FilterIncludedElements |
| ElementInclusionExclusiveFilter | - | FilterIncludedElements |
| ElementExclusionFilter | - | FilterExcludedElements |
Examples:
| # |
| # MinimumElementsFilter |
| # |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("MinimumElementsFilter") |
| # include compounds with at least 2 nitrogens and 3 oxygens |
| settingsObject[["FilterMinimumElements"]]<-"N2O3" |
| # |
| # MaximumElementsFilter |
| # |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("MaximumElementsFilter") |
| # filter out compounds with at maximum 5 nitrogens and 7 oxygens |
| settingsObject[["FilterMinimumElements"]]<-"N5O7" |
| # |
| # SmartsSubstructureInclusionFilter |
| # |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("SmartsSubstructureInclusionFilter") |
| # include compounds containing benzene |
| settingsObject[["SmartsSubstructureInclusionScoreSmartsList"]]<-c("c1ccccc1") |
| # |
| # SmartsSubstructureExclusionFilter |
| # |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("SmartsSubstructureExclusionFilter") |
| # filter out compounds containing hydroxyl groups |
| settingsObject[["SmartsSubstructureInclusionScoreSmartsList"]]<-c("[OX2H]") |
| # |
| # ElementInclusionFilter |
| # |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("ElementInclusionFilter") |
| # include compounds containing nitrogen and oxygen |
| settingsObject[["FilterIncludedElements"]]<-c("N","O") |
| # |
| # ElementExclusionFilter |
| # |
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("ElementExclusionFilter") |
| # filter out compounds including bromine or chlorine |
| settingsObject[["FilterExcludedElements"]]<-c("Cl","Br") |
Defining multiple filters at once is possible:
| settingsObject[["MetFragPreProcessingCandidateFilter"]]<-c("MinimumElementsFilter", "MaximumElementsFilter") |
| settingsObject[["FilterMinimumElements"]]<-"N2O3" |
| settingsObject[["FilterMinimumElements"]]<-"N5O7" |
Post-pocessing Candidate Filters
Filters can be defined to filter candidates after fragmentation and scoring. Following filters are available:
| InChIKeyFilter | - | filter stereoisomers by comparing first part of compounds' InChIKeys |
| only the best-scored candidate remains in the result list |
Candidate Scores
MetFrag is able to include different scores used to calculate a final score which is used to rank candidates within the candidate list. Besides pre-defined scores database dependent scores can be defined. For each additional defined score a weight needs to be defined used to calculate the final score for each candidate.
Pre-defined Candidate Scores
| FragmenterScore | - | Uses intensities, m/z values and bond energies of fragment-peak-matches |
| SmartsSubstructureInclusionScore | - | Score candiates by presence of defined substructures |
| SmartsSubstructureExclusionScore | - | Score candiates by absence of defined substructures |
| SuspectListScore | - | Score candidates by presence in defined suspect list |
| RetentionTimeScore | - | Score candidate with retention time information |
| OfflineMetFusionScore | - | Uses predifined spectral library to calculate MetFusion like similarity score |
When defining additional scores further parameters need to be defined:
| SmartsSubstructureInclusionScore | - | ScoreSmartsInclusionList |
| SmartsSubstructureExclusionScore | - | ScoreSmartsExclusionList |
| SuspectListScore | - | ScoreSuspectLists |
| - File path of file containing InChIKeys of suspect list one per line | ||
| RetentionTimeScore | - | RetentionTimeTrainingFile, ExperimentalRetentionTimeValue |
| - RetentionTimeTrainingFile is the file path of file containing retention time and InChI for logP calculation -> one per line | ||
| - example file can be found here | ||
| - mandatory columns are: RetentionTime, InChI or UserLogP | ||
| - InChI is used to calculate logP values with models included within CDK | ||
| - UserLogP (if defined) is used instead of InChI but has to be available within the candidate list as well | ||
| - RetentionTime values and ExperimentalRetentionTimeValue need to be acquired on the same system |
Database Dependent Scores
Dependent on the used database different scores. When using local file databases any score defined as candidate property can be used as scoring term.
Following scoring terms are pre-defined for available databases:
| ExtendedPubChem | - | PubChemNumberPatents,PubChemNumberPubMedReferences |
| ChemSpider | - | ChemSpiderReferenceCount,ChemSpiderNumberExternalReferences,ChemSpiderRSCCount,ChemSpiderNumberPubMedReferences,ChemSpiderDataSourceCount |
Examples
Defining FragmenterScore, SmartsSubstructureInclusionScore and RetentionTimeScore together with necessary parameters.
| settingsObject[["MetFragScoreTypes"]]<-c("FragmenterScore","SmartsSubstructureInclusionScore","RetentionTimeScore") |
| settingsObject[["ScoreSmartsInclusionList"]]<-c("[OX2H]","c1cccc1") |
| settingsObject[["RetentionTimeTrainingFile"]]<-"C:/Documents/RetentionTimeFile.csv" |
| settingsObject[["ExperimentalRetentionTimeValue"]]<-9.4 |
| settingsObject[["MetFragScoreWeights"]]<-c(1.0,0.5,0.5) |
Defining FragmenterScore and database dependent soring terms for ExtendedPubChem database.
| settingsObject[["MetFragDatabaseType"]]<-c("ExtendedPubChem") |
| settingsObject[["MetFragScoreTypes"]]<-c("FragmenterScore","PubChemNumberPatents","PubChemNumberPubMedReferences") |
| settingsObject[["MetFragScoreWeights"]]<-c(1.0,0.5,0.5) |
