pmp

pmp provides methods and tools for (pre-)processing of metabolomics datasets (i.e. peak matrices), including filtering, normalisation, missing value imputation, scaling, and signal drift and batch effect correction.

Key Features:

• Filtering Methods: Filtering based on the fraction of missing values across samples or features, Relative Standard Deviation (RSD) calculated from Quality Control (QC) samples, and blank samples.
• Normalisation Methods: Probabilistic Quotient Normalisation (PQN) and normalisation to total signal intensity.
• Missing Value Imputation: Unified interface for k-nearest neighbours (knn), random forests (rf), Bayesian PCA missing value estimator (bpca), mean or median value of the given feature, and a constant small value.
• Variance Stabilisation: Generalised logarithm (glog) transformation algorithm to stabilise variance across low and high intensity mass spectral features.
• Signal Drift and Batch Effect Correction: Implementation of the Quality Control-Robust Spline Correction (QCRSC) algorithm for mass spectrometry-based datasets.

Scientific Applications:

• Metabolomics Data Pre-processing: Comprehensive processing of peak matrices from mass spectrometry-based metabolomics experiments.
• Quality Control-Based Data Filtering: Removal of unreliable features using QC sample RSD and blank sample measurements.
• Signal Correction in Large-Scale Studies: Correction of technical variation including signal drift and batch effects in multi-batch mass spectrometry datasets.

Methodology:

The package implements filtering, normalisation, missing value imputation, glog transformation, and QCRSC algorithm for signal drift and batch effect correction of mass spectrometry-based metabolomics datasets.