rMTA: robust metabolic transformation analysis

Abstract

Motivation The development of computational tools exploiting -omics data and high-quality genome-scale metabolic networks for the identification of novel drug targets is a relevant topic in Systems Medicine. Metabolic Transformation Algorithm (MTA) is one of these tools, which aims to identify targets that transform a disease metabolic state back into a healthy state, with potential application in any disease where a clear metabolic alteration is observed. Results Here, we present a robust extension to MTA (rMTA), which additionally incorporates a worst-case scenario analysis and minimization of metabolic adjustment to evaluate the beneficial effect of gene knockouts. We show that rMTA complements MTA in the different datasets analyzed (gene knockout perturbations in different organisms, Alzheimer’s disease and prostate cancer), bringing a more accurate tool for predicting therapeutic targets. Availability and implementation rMTA is freely available on The Cobra Toolbox: https://opencobra.github.io/cobratoolbox/latest/. Supplementary information Supplementary data are available at Bioinformatics online.

Genome-scale metabolic modeling reveals SARS-CoV-2-induced metabolic changes and antiviral targets

“…As was proposed previously, targeting the virus-induced metabolic changes can be an effective antiviral strategy (Mayer et al 2019), which we adopted here to predict anti-SARS-CoV-2 targets. Specifically, we applied the GEM-based rMTA algorithm (Valcárcel et al 2019) to each of our collected datasets to predict metabolic reactions whose knockout (KO) can transform the cellular metabolism from the SARS-CoV-2-infected state to the non-infected normal state (Methods ; Table S6). MTA computes a score for each of the metabolic reactions in the cell, and usually the 10-20% reactions with the highest MTA score contain promising candidate targets (Yizhak et al 2013).…”

“…For each of the collected datasets, the DE result of virus-infected vs control samples as well as the representative flux distribution of the virus-infected group computed with iMAT (Shlomi et al 2008) followed by ACHR sampling were used as inputs for the GEM-based metabolic transformation algorithm (MTA; Yizhak et al 2013; a variant called rMTA was used; Valcárcel et al 2019) to predict metabolic reactions whose knock-out can transform cellular metabolic state from that of the virus-infected to that of the control samples (full prediction results from all datasets in Table S6A). The output of rMTA is a score (rMTA score) for each metabolic reaction, with higher scores corresponding to better candidates for achieving the metabolic transformation as specified above.…”

“…The datasets are ordered by P values, all pathways displayed have FDR<0.1. (E) The robust metabolic transformation algorithm (rMTA, Valcárcel et al 2019) was used to predict metabolic reactions whose knock-out can further transform the metabolic state of the remdesivir-treated SARS-CoV-2-infected cells back to the normal control state, using the Vero E6 cell samples (Methods). The significant metabolic pathways (FDR<0.1) enriched by the top 20% MTA-predicted targets are shown.…”

“…from diseased to healthy states). More recently, Valcárcel et al has described a variant of MTA named rMTA with improved performance (Valcárcel et al 2019). Incorporating such high-performance GEM methods in the analysis of data on SARS-CoV-2 infection provides us with a unique opportunity to understand the metabolic demands of SARS-CoV-2 and to systematically predict anti-SARS-CoV-2 targets that counteract the virus-induced metabolic alterations.…”

“…Genome-scale metabolic modeling (GEM)-based prediction of anti-SARS-CoV-2targets that act via reversing the virus-induced metabolic alterations. The robust metabolic transformation algorithm (rMTA,Valcárcel et al 2019) was used to predict metabolic reactions whose knock-out can reverse the SARS-CoV-2-induced metabolic changes using each of the collected datasets (Methods). (A) Visualization of the overlap of the top 10% MTA-predicted target reactions between each pair of datasets analyzed using Fisher's exact tests (Methods).…”