Talanta. 2022 Mar 24;244:123396. doi: 10.1016/j.talanta.2022.123396. Online ahead of print.
ABSTRACT
A computational method for the untargeted determination of cycling yeast metabolites using a comprehensive two-dimensional gas chromatography time-of-flight mass spectrometry (GC×GC-TOFMS) dataset is presented. The yeast metabolomic cycle for the diploid yeast strain CEN.PK with a 5 h cycle period relative to the O2 concentration level is comprehensively examined to determine the metabolites that exhibit cycling. Samples were collected over only two cycles (10 h with a total of 24 time-point sampling intervals at 25 min each) as an experimental constraint. Due to the limited number of cycles expressed in the dataset, a computational method was devised to determine with statistical significance whether or not a given metabolite exhibited a temporal signal pattern that constituted cycling in the context of the 5 h cycle period. The computational method we report compares the experimentally obtained 24 time-point metabolite signal sequences to randomly generated signal sequences coupled with statistically based confidence level LOF metrics to determine whether or not a given metabolite expresses cycling, and if so, what is the phase of the cycling. Initially the GC×GC-TOFMS dataset was analyzed using tile-based Fisher ratio (F-ratio) analysis. Since there were 24 time-point intervals, this constituted 24 sample classes in the F-ratio calculation which produced 672 metabolite hits. Next, application of the computational method determined that there were 210 of the 672 metabolites exhibiting cycling: 55 identified metabolites and 155 unknown metabolites. Furthermore, the 210 cycling metabolites were categorized into four groups, and where applicable, a phase determined: 1 cycle/5 h period (106 metabolites), 2 cycles/5 h period (13 metabolites), spiky pattern (12 metabolites), or multimodal pattern (79 metabolites).
PMID:35354112 | DOI:10.1016/j.talanta.2022.123396
