PCA as a Practical Indicator of OPLS-DA Model Reliability

Abstract

Background Principal Component Analysis (PCA) and Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) are powerful statistical modeling tools that provide insights into separations between experimental groups based on high-dimensional spectral measurements from NMR, MS or other analytical instrumentation. However, when used without validation, these tools may lead investigators to statistically unreliable conclusions. This danger is especially real for Partial Least Squares (PLS) and OPLS, which aggressively force separations between experimental groups. As a result, OPLS-DA is often used as an alternative method when PCA fails to expose group separation, but this practice is highly dangerous. Without rigorous validation, OPLS-DA can easily yield statistically unreliable group separation. Methods A Monte Carlo analysis of PCA group separations and OPLS-DA cross-validation metrics was performed on NMR datasets with statistically significant separations in scores-space. A linearly increasing amount of Gaussian noise was added to each data matrix followed by the construction and validation of PCA and OPLS-DA models. Results With increasing added noise, the PCA scores-space distance between groups rapidly decreased and the OPLS-DA cross-validation statistics simultaneously deteriorated. A decrease in correlation between the estimated loadings (added noise) and the true (original) loadings was also observed. While the validity of the OPLS-DA model diminished with increasing added noise, the group separation in scores-space remained basically unaffected. Conclusion Supported by the results of Monte Carlo analyses of PCA group separations and OPLS-DA cross-validation metrics, we provide practical guidelines and cross-validatory recommendations for reliable inference from PCA and OPLS-DA models.

Recent Advances in NMR-Based Metabolomics

“…Another study discussed the fact that the now-conventionally used statistical methods such as PLS-DA and OPLS-DA can mislead in the absence of proper validation and provides practical guidelines and cross-validation recommendations for reliable inference from PCA and OPLS-DA models. 65 …”

Beyond the paradigm: Combining mass spectrometry and nuclear magnetic resonance for metabolomics

“…Since the numbers of replicates for a metabolomics data set are typically far fewer then the number of variables, overfitting the data, especially for supervised techniques like PLS or OPLS, is a serious concern [65, 66, 87]. In fact, a PLS/OPLS model can produce the appearance of a clear group separation even for noise or completely random data [88].…”

“…While chemometric techniques regularly streamline the process of data analysis, these advanced multivariate statistical techniques are routinely used incorrectly, lack proper validation, and have, unfortunately, lead to a proliferation of erroneous data in the scientific literature [65, 66]. This problem becomes compounded when an investigator is combining multiple analytical sources.…”

Novel role of Tieg1 in muscle metabolism and mitochondrial oxidative capacities

“…All data sets were scaled to unit of variance allowing all metabolites to become equally important. PCA describes the total variability within the data set and can be used as an informative indicator while the PLS‐DA is used to predict the spectral features (metabolites) that define separation between groups (phenotypes) . The overall quality of the models was judged by cumulative R 2 (goodness of fit) and cumulative Q 2 (goodness of prediction).…”