DEXOM: Diversity-based enumeration of optimal context-specific metabolic networks

Abstract

The correct identification of metabolic activity in tissues or cells under different conditions can be extremely elusive due to mechanisms such as post-transcriptional modification of enzymes or different rates in protein degradation, making difficult to perform predictions on the basis of gene expression alone. Context-specific metabolic network reconstruction can overcome some of these limitations by leveraging the integration of multi-omics data into genome-scale metabolic networks (GSMN). Using the experimental information, context-specific models are reconstructed by extracting from the generic GSMN the sub-network most consistent with the data, subject to biochemical constraints. One advantage is that these context-specific models have more predictive power since they are tailored to the specific tissue, cell or condition, containing only the reactions predicted to be active in such context. However, an important limitation is that there are usually many different sub-networks that optimally fit the experimental data. This set of optimal networks represent alternative explanations of the possible metabolic state. Ignoring the set of possible solutions reduces the ability to obtain relevant information about the metabolism and may bias the interpretation of the true metabolic states. In this work we formalize the problem of enumerating optimal metabolic networks and we introduce DEXOM, an unified approach for diversity-based enumeration of context-specific metabolic networks. We developed different strategies for this purpose and we performed an exhaustive analysis using simulated and real data. In order to analyze the extent to which these results are biologically meaningful, we used the alternative solutions obtained with the different methods to measure: 1) the improvement of in silico predictions of essential genes in Saccharomyces cerevisiae using ensembles of metabolic network; and 2) the detection of alternative enriched pathways in different human cancer cell lines. We also provide DEXOM as an open-source library compatible with COBRA Toolbox 3.0, available at https://github.com/MetExplore/dexom.

Transcriptomics data integration for context-specific modeling of Atlantic salmon metabolism: functional evaluation of methods based on metabolic tasks

“…There are many other MEMs available that were not systematically tested in this or other studies, including recent methods that account for trancriptomic variability (Joshi et al, 2020), use ensemble modeling to improve predictions (Rodríguez-Mier et al, 2021), or combine multiple methods and settings (Vieira et al, 2022). Notably, some methods integrate metabolic tasks into the model extraction procedure itself by requiring agreement with inferred task feasibility for human models and data (Agren et al, 2014;Richelle et al, 2019b).…”

“…Ensemble modeling could also help improve the models themselves by applying machine learning to their contents and predictions (Medlock and Papin, 2020). Indeed, recent studies have demonstrated context-specific ensemble modeling with a single MEM (Rodríguez-Mier et al, 2021) and combined multiple MEMs to build a single model (Vieira et al, 2022).…”

Extracting functionally accurate context-specific models of Atlantic salmon metabolism

“…Ensemble modeling could also help improve the models themselves by applying machine learning to their contents and predictions [48]. Indeed, recent studies have demonstrated context-specific ensemble modeling with a single MEM [49] and combined multiple MEMs to build a single model [50].…”

Multi-Omics Integrative Analysis Coupled to Control Theory and Computational Simulation of a Genome-Scale metabolic Model Reveal Controlling Biological Switches in Human Astrocytes Under Palmitic Acid-Induced Lipotoxicity

“…The methodology allows the creation of not correlated PCs, letting the information of linearly correlated variables to be represented in one PC. These permit the construction of an estimated biologically significant variable (Rodríguez-Mier et al, 2021) and provides a way to limit the potential for overfitting (Wörheide et al, 2021).…”