# Multivariate Data Analysis Methods and Their Application in Lipidomics: A Gentle Comment on Appropriateness and Reliability Criteria

## TL;DR

The paper defends the use of PCA and PLS-DA in lipidomics, emphasizing their reliability and appropriateness when applied rigorously.

## Contribution

It clarifies the methodological validity of PCA and PLS-DA in lipidomics against recent critiques and reaffirms their standard use.

## Key findings

- PCA and PLS-DA are suitable for capturing variations in lipidomics data from controlled in vitro models.
- These methods are robust when combined with statistical testing and biological validation.
- Linear models remain reliable even with the availability of nonlinear techniques.

## Abstract

In response to Yoshiyasu Takefuji's critique regarding the use of Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS‐DA) in the study “Melatonin Repairs the Lipidome of Human Hepatocytes Exposed to Cd and Free Fatty Acid‐Induced Lipotoxicity,” we provide a methodological clarification. PCA and PLS‐DA are well‐established, widely validated tools for exploratory analysis of high‐dimensional omics data, including lipidomics data. Although these methods are linear, they are appropriate for capturing systematic and directional variations in complex biological systems, particularly in controlled in vitro models like ours. Our analytical approach integrates PCA and PLS‐DA with rigorous statistical testing, data transformations, and biological validation, ensuring robustness and biological relevance of the findings. We reaffirm that these methods represent a standard, reliable practice in lipidomics, and the potential of nonlinear techniques does not diminish the appropriateness or utility of linear multivariate models when applied with scientific rigor.