Minimum variance optimized Fisher ratio analysis of comprehensive two-dimensional gas chromatography / mass spectrometry data: Study of the pacu fish metabolome

Optimization of Parameters for ROI Data Compression for Nontargeted Analyses Using LC–HRMS

“…While this was visualized tuning only one parameter, it is advised to optimize both parameters simultaneously with spiked standards as in Figure to ensure that the data are accurately represented for the intended analysis, especially for low-concentration analytes of interest. Furthermore, we expect this workflow to be readily adapted to comprehensive two-dimensional LC (LC × LC) with HRMS data as both the ROI procedure and tile-based F -ratio analysis have been demonstrated for various comprehensive two-dimensional chromatography data applications. ,,,− …”

“…All data were imported to Matlab 2020b (MathWorks Inc., Natick, MA, USA), where it was processed using the ROI procedure ,, prior to performing in-house tile-based F -ratio analysis. ,, The ROI workflow is presented in Figure , with a schematic describing each step of the data processing prior to F -ratio analysis. This workflow requires no additional preprocessing and effectively compresses HRMS data without losing the detector mass accuracy.…”

“…Using HRMS increases the probability of obtaining pure m/z per analyte as the mass accuracy improves, though not all m/z necessarily could pass the signal threshold to be considered by F -ratio analysis. To improve discovery of true positive hits, a recent development was applied, herein termed the minimum variance optimized (MVO) F -ratio ( MVO F -ratio) analysis, replacing the denominator with the minimum within-class variance. For every tile, the within-class variance is calculated for each class, and the lowest variance is used.…”

Recent Advances in GC×GC and Chemometrics to Address Emerging Challenges in Nontargeted Analysis

“…The analysis on the effects of an integrated rice-farming system on the pacu fish (Piaractus mesopotamicus) metabolome benefited from another modification to the F-ratio metric, termed minimum variance optimized FRA ( MVO FRA). 73 This method modifies the F-ratio calculation on a per-tile basis, where the class with the lowest variance in a given tile is used in the denominator. This ensures significant analytes are not overlooked when the variance for a particular class is not uniformly low or high.…”

Tile-Based Random Forest Analysis for Analyte Discovery in Balanced and Unbalanced GC × GC-TOFMS Data Sets

“…15−18 Nonetheless, several research studies have proposed some alterations to the aforementioned method to expand its scope to better suit specific practical settings. In this respect, multiple F-ratio amended calculations were suggested, such as the controlnormalized F-ratio 19 ( CN FRA) and the minimum variance optimized F-ratio 20 ( MVO FRA). Furthermore, multiple studies investigated combining the tiling approach with other metrics, such as relative variance (RSD 2 ) in the tile-based variance rank initiated-unsupervised sample indexing (VRI-USI) method 21 and the use of a rank metric (RM) for pairwise analysis workflows.…”