Computational drug repositioning identifies niclosamide and tribromsalan as inhibitors of Mycobacterium tuberculosis and Mycobacterium abscessus.

Tuberculosis (TB) is still a major global health challenge, killing over 1.5 million people each year, and hence, there is a need to identify and develop novel treatments for Mycobacterium tuberculosis (M. tuberculosis). The prevalence of infections caused by nontuberculous mycobacteria (NTM) is also increasing and has overtaken TB cases in the United States and much of the developed world. Mycobacterium abscessus (M. abscessus) is one of the most frequently encountered NTM and is difficult to treat. We describe the use of drug-disease association using a semantic knowledge graph approach combined with machine learning models that has enabled the identification of several molecules for testing anti-mycobacterial activity. We established that niclosamide (M. tuberculosis IC90 2.95 µM; M. abscessus IC90 59.1 µM) and tribromsalan (M. tuberculosis IC90 76.92 µM; M. abscessus IC90 147.4 µM) inhibit M. tuberculosis and M. abscessus in vitro. To investigate the mode of action, we determined the transcriptional response of M. tuberculosis and M. abscessus to both compounds in axenic log phase, demonstrating a broad effect on gene expression that differed from known M. tuberculosis inhibitors. Both compounds elicited transcriptional responses indicative of respiratory pathway stress and the dysregulation of fatty acid metabolism.

Genome-Scale Metabolic Modeling Enables In-Depth Understanding of Big Data.

Genome-scale metabolic models (GEMs) enable the mathematical simulation of the metabolism of archaea, bacteria, and eukaryotic organisms. GEMs quantitatively define a relationship between genotype and phenotype by contextualizing different types of Big Data (e.g., genomics, metabolomics, and transcriptomics). In this review, we analyze the available Big Data useful for metabolic modeling and compile the available GEM reconstruction tools that integrate Big Data. We also discuss recent applications in industry and research that include predicting phenotypes, elucidating metabolic pathways, producing industry-relevant chemicals, identifying drug targets, and generating knowledge to better understand host-associated diseases. In addition to the up-to-date review of GEMs currently available, we assessed a plethora of tools for developing new GEMs that include macromolecular expression and dynamic resolution. Finally, we provide a perspective in emerging areas, such as annotation, data managing, and machine learning, in which GEMs will play a key role in the further utilization of Big Data.

Modeling of nitrogen fixation and polymer production in the heterotrophic diazotroph Azotobacter vinelandii DJ.

Nitrogen fixation is an important metabolic process carried out by microorganisms, which converts molecular nitrogen into inorganic nitrogenous compounds such as ammonia (NH3). These nitrogenous compounds are crucial for biogeochemical cycles and for the synthesis of essential biomolecules, i.e. nucleic acids, amino acids and proteins. Azotobacter vinelandii is a bacterial non-photosynthetic model organism to study aerobic nitrogen fixation (diazotrophy) and hydrogen production. Moreover, the diazotroph can produce biopolymers like alginate and polyhydroxybutyrate (PHB) that have important industrial applications. However, many metabolic processes such as partitioning of carbon and nitrogen metabolism in A. vinelandii remain unknown to date. Genome-scale metabolic models (M-models) represent reliable tools to unravel and optimize metabolic functions at genome-scale. M-models are mathematical representations that contain information about genes, reactions, metabolites and their associations. M-models can simulate optimal reaction fluxes under a wide variety of conditions using experimentally determined constraints. Here we report on the development of a M-model of the wild type bacterium A. vinelandii DJ (iDT1278) which consists of 2,003 metabolites, 2,469 reactions, and 1,278 genes. We validated the model using high-throughput phenotypic and physiological data, testing 180 carbon sources and 95 nitrogen sources. iDT1278 was able to achieve an accuracy of 89% and 91% for growth with carbon sources and nitrogen source, respectively. This comprehensive M-model will help to comprehend metabolic processes associated with nitrogen fixation, ammonium assimilation, and production of organic nitrogen in an environmentally important microorganism.

RepTB: a gene ontology based drug repurposing approach for tuberculosis.

Tuberculosis (TB) is the world's leading infectious killer with 1.8 million deaths in 2015 as reported by WHO. It is therefore imperative that alternate routes of identification of novel anti-TB compounds are explored given the time and costs involved in new drug discovery process. Towards this, we have developed RepTB. This is a unique drug repurposing approach for TB that uses molecular function correlations among known drug-target pairs to predict novel drug-target interactions. In this study, we have created a Gene Ontology based network containing 26,404 edges, 6630 drug and 4083 target nodes. The network, enriched with molecular function ontology, was analyzed using Network Based Inference (NBI). The association scores computed from NBI are used to identify novel drug-target interactions. These interactions are further evaluated based on a combined evidence approach for identification of potential drug repurposing candidates. In this approach, targets which have no known variation in clinical isolates, no human homologs, and are essential for Mtb's survival and or virulence are prioritized. We analyzed predicted DTIs to identify target pairs whose predicted drugs may have synergistic bactericidal effect. From the list of predicted DTIs from RepTB, four TB targets, namely, FolP1 (Dihydropteroate synthase), Tmk (Thymidylate kinase), Dut (Deoxyuridine 5'-triphosphate nucleotidohydrolase) and MenB (1,4-dihydroxy-2-naphthoyl-CoA synthase) may be selected for further validation. In addition, we observed that in some cases there is significant chemical structure similarity between predicted and reported drugs of prioritized targets, lending credence to our approach. We also report new chemical space for prioritized targets that may be tested further. We believe that with increasing drug-target interaction dataset RepTB will be able to offer better predictive value and is amenable for identification of drug-repurposing candidates for other disease indications too.

Assessment of Right Ventricular Function by Newer Imaging in Echocardiography in Idiopathic Pulmonary Arterial Hypertension.

BACKGROUND: The aims of the study were to assess the right ventricular (RV) functions in patients with idiopathic pulmonary arterial hypertension (IPAH) with RV longitudinal strain (RVLS) in addition to conventional parameters, as well as its correlation with severity and prognosis in IPAH. METHODS: Twenty-two IPAH patients were followed up for 1 year. ANOVA and Gabriel's pairwise comparison tests were used for comparison of RVLS with respect to WHO functional class status. Patients were divided into non-survival (group 1) and survival (group 2), and clinical and echocardiographic parameters of RV function were compared at baseline and at 6 months with t-test & Mann-Whitney test. RESULTS: At baseline, with respect to WHO functional class, mean RVLS showed no significant interclass difference (P = 0.0781). Among the other conventional echocardiographic parameters, RV E/A showed significant difference at baseline (P = 0.004), but not at 6 months (P = 0.366); whereas tricuspid annular plane systolic excursion (TAPSE) which had no significant difference initially (P = 0.174) revealed a significance level at 6 months (P = 0.029) between the two groups. Fractional area change (FAC), RV index of myocardial performance (RIMP), and right atrial (RA) area displayed significant difference neither at baseline nor at 6 months. RVLS exhibited significant difference neither at baseline (P = 0.912) nor at 6 months (P = 0.181). None of the echocardiographic parameters including RVLS showed a significant average change with change in severity of PAH both at 6 and 12 months. CONCLUSION: RVLS was not proved to be a useful parameter for early detection of RV dysfunction and prognosis in patients with IPAH in comparison with the conventional echocardiographic parameters.

Crowd sourcing a new paradigm for interactome driven drug target identification in Mycobacterium tuberculosis.

A decade since the availability of Mycobacterium tuberculosis (Mtb) genome sequence, no promising drug has seen the light of the day. This not only indicates the challenges in discovering new drugs but also suggests a gap in our current understanding of Mtb biology. We attempt to bridge this gap by carrying out extensive re-annotation and constructing a systems level protein interaction map of Mtb with an objective of finding novel drug target candidates. Towards this, we synergized crowd sourcing and social networking methods through an initiative 'Connect to Decode' (C2D) to generate the first and largest manually curated interactome of Mtb termed 'interactome pathway' (IPW), encompassing a total of 1434 proteins connected through 2575 functional relationships. Interactions leading to gene regulation, signal transduction, metabolism, structural complex formation have been catalogued. In the process, we have functionally annotated 87% of the Mtb genome in context of gene products. We further combine IPW with STRING based network to report central proteins, which may be assessed as potential drug targets for development of drugs with least possible side effects. The fact that five of the 17 predicted drug targets are already experimentally validated either genetically or biochemically lends credence to our unique approach.

In-silico predictive mutagenicity model generation using supervised learning approaches.

BACKGROUND: Experimental screening of chemical compounds for biological activity is a time consuming and expensive practice. In silico predictive models permit inexpensive, rapid "virtual screening" to prioritize selection of compounds for experimental testing. Both experimental and in silico screening can be used to test compounds for desirable or undesirable properties. Prior work on prediction of mutagenicity has primarily involved identification of toxicophores rather than whole-molecule predictive models. In this work, we examined a range of in silico predictive classification models for prediction of mutagenic properties of compounds, including methods such as J48 and SMO which have not previously been widely applied in cheminformatics. RESULTS: The Bursi mutagenicity data set containing 4337 compounds (Set 1) and a Benchmark data set of 6512 compounds (Set 2) were taken as input data set in this work. A third data set (Set 3) was prepared by joining up the previous two sets. Classification algorithms including Naïve Bayes, Random Forest, J48 and SMO with 10 fold cross-validation and default parameters were used for model generation on these data sets. Models built using the combined performed better than those developed from the Benchmark data set. Significantly, Random Forest outperformed other classifiers for all the data sets, especially for Set 3 with 89.27% accuracy, 89% precision and ROC of 95.3%. To validate the developed models two external data sets, AID1189 and AID1194, with mutagenicity data were tested showing 62% accuracy with 67% precision and 65% ROC area and 91% accuracy, 91% precision with 96.3% ROC area respectively. A Random Forest model was used on approved drugs from DrugBank and metabolites from the Zinc Database with True Positives rate almost 85% showing the robustness of the model. CONCLUSION: We have created a new mutagenicity benchmark data set with around 8,000 compounds. Our work shows that highly accurate predictive mutagenicity models can be built using machine learning methods based on chemical descriptors and trained using this set, and these models provide a complement to toxicophores based methods. Further, our work supports other recent literature in showing that Random Forest models generally outperform other comparable machine learning methods for this kind of application.