| Popis: |
Integration of rice and fish farming, eg., pacu fish in Argentina, has raised concern that herbicides used for rice paddies may adversely affect the fish metabolome. To study this issue, tile-based Fisher ratio (F-ratio) analysis was applied to comprehensive two-dimensional gas chromatography with time-of-flight mass spectrometry (GC × GC-TOFMS) data of pacu fish raised in an integrated rice-fish farming system (farmed class) versus fish raised in tanks (tank-raised class) to discover class-distinguishing analytes. F-ratio analysis resulted in hit lists initially dominated by artifact peaks from the sample derivatization process, as well as some redundant hits. These challenges were addressed by developing an automated artifact removal algorithm and an improvement to redundant hit removal in the tile-based F-ratio analysis using a pooled farmed fish sample, either spiked with 29 metabolites (spiked class) or unspiked (i.e., the background serving as the control class). Of the 29 spiked metabolites, 23 were discovered by standard F-ratio analysis, improving to 28 discovered using control-normalized F-ratio analysis. Standard and control-normalized F-ratio hit lists initially with 185 and 246 hits, were reduced to 56 and 49 hits, respectively, after artifact removal and removing redundant hits. Next, we returned to the F-ratio analysis of the farmed fish versus the control fish. Here, we introduce a minimum variance optimized (MVO) F-ratio calculation ( |
Full Text from ScienceDirect