KODAMA exploratory analysis in metabolic phenotyping

Abstract

KODAMA is a valuable tool in metabolomics research to perform exploratory analysis. The advanced analytical technologies commonly used for metabolic phenotyping, mass spectrometry, and nuclear magnetic resonance spectroscopy push out a bunch of high-dimensional data. These complex datasets necessitate tailored statistical analysis able to highlight potentially interesting patterns from a noisy background. Hence, the visualization of metabolomics data for exploratory analysis revolves around dimensionality reduction. KODAMA excels at revealing local structures in high-dimensional data, such as metabolomics data. KODAMA has a high capacity to detect different underlying relationships in experimental datasets and correlate extracted features with accompanying metadata. Here, we describe the main application of KODAMA exploratory analysis in metabolomics research.

Keywords: KODAMA; clustering; metabolomics; semi-supervised; unsupervised.

FIGURE 1

KODAMA on machine learning algorithm’s map. The machine learning algorithm can be categorized in unsupervised (i.e., clustering and dimensionality reduction method) and supervised learning. KODAMA is one of unsupervised learning methods used for dimensionality reduction. Optionally, if supervised information is used to lead the process of discovery of new patterns, KODAMA can be classified as semi-supervised.

FIGURE 2

KODAMA accuracy maximization by iterative cross-validations. (A) Cross-validation model (CV) generates predicted labels (PLs) that are used to calculate the accuracy value (AC). (B) Generation of new labels to conduct the process of accuracy maximization can be i) an unsupervised method, randomly swapping some class labels of misleading samples with predicted labels; ii) semi-supervised type-I, changing only predefined labels and maintaining assigned class labels; or iii) semi-supervised type-II, changing groups of labels together forcing their belonging to the same class. (C) Generation of new labels is an iterative process aimed to identify the labels with the highest cross-validated accuracy. (D) Accuracy values increase with the number of iterations. (E) KODAMA dissimilarity matrix generated as output can be transformed with MDS, or t-SNE, in a low-dimensional space.