WE-ASCA: The Weighted-Effect ASCA for Analyzing Unbalanced Multifactorial Designs—A Raman Spectra-Based Example

Abstract

Analyses of multifactorial experimental designs are used as an explorative technique describing hypothesized multifactorial effects based on their variation. The procedure of analyzing multifactorial designs is well established for univariate data, and it is known as analysis of variance (ANOVA) tests, whereas only a few methods have been developed for multivariate data. In this work, we present the weighted-effect ASCA, named WE-ASCA, as an enhanced version of ANOVA-simultaneous component analysis (ASCA) to deal with multivariate data in unbalanced multifactorial designs. The core of our work is to use general linear models (GLMs) in decomposing the response matrix into a design matrix and a parameter matrix, while the main improvement in WE-ASCA is to implement the weighted-effect (WE) coding in the design matrix. This WE-coding introduces a unique solution to solve GLMs and satisfies a constrain in which the sum of all level effects of a categorical variable equal to zero. To assess the WE-ASCA performance, two applications were demonstrated using a biomedical Raman spectral data set consisting of mice colorectal tissue. The results revealed that WE-ASCA is ideally suitable for analyzing unbalanced designs. Furthermore, if WE-ASCA is applied as a preprocessing tool, the classification performance and its reproducibility can significantly improve.

CARS Imaging Advances Early Diagnosis of Cardiac Manifestation of Fabry Disease

“…The resulting constructed model was then utilized to identify the GLA KO tissues in the acquired datasets. In this context, similar PCA-LDA models were successfully applied for a supervised prediction of tumor tissue [55,56]. Here, the PCA-LDA models were conducted on the mean spectra of tissue sections without image scaling using an increasing number of principal components (PCs).…”

Multivariate analysis of NMR‐based metabolomic data

“…69 It can also be used as a pre-processing step, to remove undesired variation from the data before training classification models. 70 As long as the design is balanced and the proper constraints are put on the model (ie using classical ANOVA effect coding), the effects described by the different effect matrices will be fully orthogonal. In this situation it is possible to quantify and decompose the sums of squares from each factors and thereafter calculate the explained variance for a given factor.…”

“…For example, it has been widely used in the field of nutritional metabolomics to study the effect of nutritional interventions, 67 but is also seeing increasing use in human clinical trials, 68 as well as in observational settings 69 . It can also be used as a pre‐processing step, to remove undesired variation from the data before training classification models 70 …”

A tutorial on the analysis of multifactorial designs from one or more data sources using AComDim

“…To tackle this issue, Thiel et al 22 suggested a formulation of ANOVA models based on the GLM methodology, which decomposes the response matrix using a factor coding called sum coding, but other strategies for the coding of factors also exist. 23 More recently, weighted-effect (WE) coding 24,25 was combined with ASCA 26 as a type of coding in order to handle unbalanced designs. In the end, all these methodologies serve the same purpose, which is to best estimate the effect matrices by decomposing the data matrix X as described in Equation (2).…”