Pinpointing Cancer Sub-Type Specific Metabolic Tasks Facilitates Identification of Anti-cancer Targets.

Metabolic reprogramming is one of the hallmarks of tumorigenesis. Understanding the metabolic changes in cancer cells may provide attractive therapeutic targets and new strategies for cancer therapy. The metabolic states are not the same in different cancer types or subtypes, even within the same sample of solid tumors. In order to understand the heterogeneity of cancer cells, we used the Pareto tasks inference method to analyze the metabolic tasks of different cancers, including breast cancer, lung cancer, digestive organ cancer, digestive tract cancer, and reproductive cancer. We found that cancer subtypes haves different propensities toward metabolic tasks, and the biological significance of these metabolic tasks also varies greatly. Normal cells treat metabolic tasks uniformly, while different cancer cells focus on different pathways. We then integrated the metabolic tasks into the multi-objective genome-scale metabolic network model, which shows higher accuracy in the in silico prediction of cell states after gene knockout than the conventional biomass maximization model. The predicted potential single drug targets could potentially turn into biomarkers or drug design targets. We further implemented the multi-objective genome-scale metabolic network model to predict synthetic lethal target pairs of the Basal and Luminal B subtypes of breast cancer. By analyzing the predicted synthetic lethal targets, we found that mitochondrial enzymes are potential targets for drug combinations. Our study quantitatively analyzes the metabolic tasks of cancer and establishes cancer type-specific metabolic models, which opens a new window for the development of specific anti-cancer drugs and provides promising treatment plans for specific cancer subtypes.

Identification of Cancer-associated metabolic vulnerabilities by modeling multi-objective optimality in metabolism.

BACKGROUND: Cancer cells undergo global reprogramming of cellular metabolism to satisfy demands of energy and biomass during proliferation and metastasis. Computational modeling of genome-scale metabolic models is an effective approach for designing new therapeutics targeting dysregulated cancer metabolism by identifying metabolic enzymes crucial for satisfying metabolic goals of cancer cells, but nearly all previous studies neglect the existence of metabolic demands other than biomass synthesis and trade-offs between these contradicting metabolic demands. It is thus necessary to develop computational models covering multiple metabolic objectives to study cancer metabolism and identify novel metabolic targets. METHODS: We developed a multi-objective optimization model for cancer cell metabolism at genome-scale and an integrated, data-driven workflow for analyzing the Pareto optimality of this model in achieving multiple metabolic goals and identifying metabolic enzymes crucial for maintaining cancer-associated metabolic phenotypes. Using this workflow, we constructed cell line-specific models for a panel of cancer cell lines and identified lists of metabolic targets promoting or suppressing cancer cell proliferation or the Warburg Effect. The targets were then validated using knockdown and over-expression experiments in cultured cancer cell lines. RESULTS: We found that the multi-objective optimization model correctly predicted phenotypes including cell growth rates, essentiality of metabolic genes and cell line specific sensitivities to metabolic perturbations. To our surprise, metabolic enzymes promoting proliferation substantially overlapped with those suppressing the Warburg Effect, suggesting that simply targeting the overlapping enzymes may lead to complicated outcomes. We also identified lists of metabolic enzymes important for maintaining rapid proliferation or high Warburg Effect while having little effect on the other. The importance of these enzymes in cancer metabolism predicted by the model was validated by their association with cancer patient survival and knockdown and overexpression experiments in a variety of cancer cell lines. CONCLUSIONS: These results confirm this multi-objective optimization model as a novel and effective approach for studying trade-off between metabolic demands of cancer cells and identifying cancer-associated metabolic vulnerabilities, and suggest novel metabolic targets for cancer treatment.

CavityPlus: a web server for protein cavity detection with pharmacophore modelling, allosteric site identification and covalent ligand binding ability prediction.

CavityPlus is a web server that offers protein cavity detection and various functional analyses. Using protein three-dimensional structural information as the input, CavityPlus applies CAVITY to detect potential binding sites on the surface of a given protein structure and rank them based on ligandability and druggability scores. These potential binding sites can be further analysed using three submodules, CavPharmer, CorrSite, and CovCys. CavPharmer uses a receptor-based pharmacophore modelling program, Pocket, to automatically extract pharmacophore features within cavities. CorrSite identifies potential allosteric ligand-binding sites based on motion correlation analyses between cavities. CovCys automatically detects druggable cysteine residues, which is especially useful to identify novel binding sites for designing covalent allosteric ligands. Overall, CavityPlus provides an integrated platform for analysing comprehensive properties of protein binding cavities. Such analyses are useful for many aspects of drug design and discovery, including target selection and identification, virtual screening, de novo drug design, and allosteric and covalent-binding drug design. The CavityPlus web server is freely available at http://repharma.pku.edu.cn/cavityplus or http://www.pkumdl.cn/cavityplus.

Deep Learning for Drug-Induced Liver Injury.

Drug-induced liver injury (DILI) has been the single most frequent cause of safety-related drug marketing withdrawals for the past 50 years. Recently, deep learning (DL) has been successfully applied in many fields due to its exceptional and automatic learning ability. In this study, DILI prediction models were developed using DL architectures, and the best model trained on 475 drugs predicted an external validation set of 198 drugs with an accuracy of 86.9%, sensitivity of 82.5%, specificity of 92.9%, and area under the curve of 0.955, which is better than the performance of previously described DILI prediction models. Furthermore, with deep analysis, we also identified important molecular features that are related to DILI. Such DL models could improve the prediction of DILI risk in humans. The DL DILI prediction models are freely available at http://www.repharma.cn/DILIserver/DILI_home.php.