Skip to contents

Impute missing variables

Source: R/metabolomics_computation.R
met.impute.Rd

Replace missing variables via a chosen method. Data need to be re-calibrated after this step, including met.PerformFeatureFilter as well as met.normalize. Data imputation is performed as part of the data preparation workflow met.read_data.

Usage

met.impute(mSetObj = NA, method = "min")

Arguments

mSetObj

Enter the name of the created mSet object ((see met.initialize and Read.TextData).

method

(Character) Select the option to replace missing variables:

  • "lod" replaces missing values with 1/5 of the minimum value for the respective variable.

  • "rowmin" replaces missing values with the half sample minimum.

  • "colmin" replaces missing values with the half feature minimum.

  • "mean" replaces missing values with the mean value of the respective feature column.

  • "median" replaces missing values with the median value of the respective feature column.

  • "knn_var" imputes missing values by finding the features in the training set “closest” to it and averages these nearby points to fill in the value.

  • "knn_smp" imputes missing values by finding the samples in the training set “closest” to it and averages these nearby points to fill in the value.

  • "bpca" applies Bayesian PCA to impute missing values.

  • "ppca" applies probabilistic PCA to impute missing values.

  • "svdImpute" applies singular value decomposition to impute missing values.

Value

The input mSet object with imputed data at mSetObj$dataSet$data_proc.

References

adapted from ImputeMissingVar (https://github.com/xia-lab/MetaboAnalystR).

Author

Nicolas T. Wirth mail.nicowirth@gmail.com Technical University of Denmark License: GNU GPL (>= 2)