README.md

In kimsche/MetaboMate: All you need for NMR-based metabolic profiling with R!

MetaboMate

All you need for NMR-based metabolic profiling with R!

Background

The MetaboMate package is the result of my PhD studies and PostDoc time at Imperial College London, where I performed metabolic analysis of urine and plasma samples using Bruker IVDr methods. MetaboMate provides R-tools covering the NMR data analaysis pipeline, including data import, pre-processing and qunatitative comparison using multivariate statistical techniques. All scripts were developed by myself (Torben Kimhofer) and improved over the years. Please feel free to get in touch with me for any queries ([e-mail]).

The MetaboMate project is discontinued from August 2020 onwards

The trend of using larger study cohorts in the field of metabolomics has motivated me to develop a follow-up package: metabom8. This successor package speeds up computation times and the improvement in run-time is achieved by implementing key functions in C++ using the Rcpp package. This means that the MetaboMate project is discontinued and not any longer supported from August 2020 onwards. The new time-improved package (metabom8) differs in R object definitions and also includes additional functionalities for improved (and fast) statistical model building, metabolite identification and data visualisation. metabom8 will be linked to this website and released in due time. -TK

The MetaboMate R toolbox covers all basic data pre-processing and analysis steps, including:

• Data Import
  • Fast read-in of 1D NMR spectra (Bruker NMR)
  • Import of spectrometer parameters (defined by Topspin)
    • Acquisition parameters, eg. spectral frequency, AU program, receiver gain)
    • Processing details, eg. FID processing details such as type of apodisation function or zerofilling paramters
• Spectral Quality Assessment
  • Detection of baseline distortions
  • Quality assessment of water suppression
  • Artifact detection arising from inadequate shimming and phasing procedures
  • Signal-to-noise estimation
• Data Pre-Processing
  • Calibration to external standard (e.g., TSP or glucose signal),
  • Baseline correction,
  • User-friendly excision of spectral ranges (eg. residual water)
• Sample Normalisation
  • Total area normalisation
  • Probablistic quotient normalisation (PQN)
  • Evaluation of sample dilution coefficients
• Multivariate Analysis and Statistics
  • Principal Component Analysis (PCA)
  • Orthogoanal-Patial Least Squares (O-PLS) regression (R) and discriminant analysis (DA)
  • Automatic selection of the optimal number of orthogonal components (O-PLS-DA)
  • Q2 and cross-validated Area Under ROC curve indices to estimate O-PLS model fits (O-PLS-R and DA, respectively)
  • Distance to Model in X space for detection of moderate outliers
  • Hotelling's T2 statistic
  • Single or multi-column Y for O-PLS

Statistical model interpretation is supported by aestethically pleasing visualisations using ggplot2.

The provided functions to perform statistics (PCA nd O-PLS) can also be applied to mass spectrometry data.

All functions are fully documented, assessible with the help() function in the R console.

kimsche/MetaboMate documentation built on Aug. 8, 2020, 1:14 a.m.