Chemometric approach to fatty acid metabolism-distribution networks and methane production in ruminal microbiome

• Methane emission has attracted more attention from nutritional and environmental scientists.
• We can develop a Chemometrics methodology to integrate different parallel laboratory experiments.
• We can approach to experimental data of methane production with perturbations theory.
• These perturbation theory models are useful to assemble fatty acid distribution networks.

Methane emission has been attracting more and more attention. Unfortunately, a lot of factors influence methane emission (chemical structure of metabolites, time, methane, gas pressure, microbiome composition, diet, etc.). We propose a new chemometric methodology to integrate different laboratory experiments in this field. Firstly, we report (1) new laboratory experiments to measure by separating (1a) methane production (gas phase), (1b) volatile fatty acid (VFA) distribution (liquid phase) and (1c) fatty acid (FA) distribution in rumen microbiome. Next, we also report a new (2) chemometric methodology that integrates all the data in a single theoretical model. The laboratory work includes two experimental sections (a) to measure the methane production, pH, gas pressure, temperature and (b) FA distribution. Section (b) includes two different experimental parts: chromatographic determination of internal peak areas (IPA%) of (b.1) long-chain fatty acids (LCFA) and (b.2) VFA. In all studies, we can use different treatments, distribution phases (media, bacteria, or protozoan microbiome), cis/trans patterns, experimental protocols, etc. Next, we combined perturbation theory (PT), linear free-energy relationships (LFER), linear discriminant analysis (LDA), and artificial neural networks (ANNs) to develop linear and non-linear models of perturbations in methane production–fatty acid distribution network. The best PT-LFER model found presented values of sensitivity, specificity, and accuracy > 0.94, and Matthews correlation coefficient (MCC) > 0.894 for 545,695 cases of perturbations in experimental data. This methodology may be useful to quantify the effect of perturbations due to the changes in experimental conditions in the study of fatty acid distribution when we need to carry out parallel experiments in different phases.