Abstract: MOTIVATION: Computational drug repositioning is a cost-effective strategy to identify novel indications for existing drugs. Drug repositioning is often modeled as a recommendation system problem. Taking advantage of the known drug-disease associations, the objective of the recommendation system is to identify new treatments by filling out the unknown entries in the drug-disease association matrix, which is known as matrix completion. Underpinned by the fact that common molecular pathways contribute to many different diseases, the recommendation system assumes that the underlying latent factors determining drug-disease associations are highly correlated. In other words, the drug-disease matrix to be completed is low-rank. Accordingly, matrix completion algorithms efficiently constructing low-rank drug-disease matrix approximations consistent with known associations can be of immense help in discovering the novel drug-disease associations. RESULTS: In this article, we propose to use a bounded nuclear norm regularization (BNNR) method to complete the drug-disease matrix under the low-rank assumption. Instead of strictly fitting the known elements, BNNR is designed to tolerate the noisy drug-drug and disease-disease similarities by incorporating a regularization term to balance the approximation error and the rank properties. Moreover, additional constraints are incorporated into BNNR to ensure that all predicted matrix entry values are within the specific interval. BNNR is carried out on an adjacency matrix of a heterogeneous drug-disease network, which integrates the drug-drug, drug-disease and disease-disease networks. It not only makes full use of available drugs, diseases and their association information, but also is capable of dealing with cold start naturally. Our computational results show that BNNR yields higher drug-disease association prediction accuracy than the current state-of-the-art methods. The most significant gain is in prediction precision measured as the fraction of the positive predictions that are truly positive, which is particularly useful in drug design practice. Cases studies also confirm the accuracy and reliability of BNNR. AVAILABILITY AND IMPLEMENTATION: The code of BNNR is freely available at https://github.com/BioinformaticsCSU/BNNR. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Abstract: Induction of specific cellular transitions is of clinical importance, as it allows to revert disease cellular phenotype, or induce cellular reprogramming and differentiation for regenerative medicine. Signalling is a convenient way to accomplish such transitions without transfer of genetic material. Here we present the first general computational method that systematically predicts signalling molecules, whose perturbations induce desired cellular transitions. This probabilistic method integrates gene regulatory networks (GRNs) with manually-curated signalling pathways obtained from MetaCore from Clarivate Analytics, to model how signalling cues are received and processed in the GRN. The method was applied to 219 cellular transition examples, including cell type transitions, and overall correctly predicted experimentally validated signalling molecules, consistently outperforming other well-established approaches, such as differential gene expression and pathway enrichment analyses. Further, we validated our method predictions in the case of rat cirrhotic liver, and identified the activation of angiopoietins receptor Tie2 as a potential target for reverting the disease phenotype. Experimental results indicated that this perturbation induced desired changes in the gene expression of key TFs involved in fibrosis and angiogenesis. Importantly, this method only requires gene expression data of the initial and desired cell states, and therefore is suited for the discovery of signalling interventions for disease treatments and cellular therapies.

Abstract: The protein-protein interaction (PPI) network of an organism serves as a skeleton for its signaling circuitry, which mediates cellular response to environmental and genetic cues. Understanding this circuitry could improve the prediction of gene function and cellular behavior in response to diverse signals. To realize this potential, one has to comprehensively map PPIs and their directions of signal flow. While the quality and the volume of identified human PPIs improved dramatically over the last decade, the directions of these interactions are still mostly unknown, thus precluding subsequent prediction and modeling efforts. Here we present a systematic approach to orient the human PPI network using drug response and cancer genomic data. We provide a diffusion-based method for the orientation task that significantly outperforms existing methods. The oriented network leads to improved prioritization of cancer driver genes and drug targets compared to the state-of-the-art unoriented network.

Abstract: Biological target (commonly genes or proteins) identification is still largely a manual process, where experts manually try to collect and combine information from hundreds of data sources, ranging from scientific publications to omics databases. Targeting the wrong gene or protein will lead to failure of the drug development process, as well as incur delays and costs. To improve this process, different software platforms are being developed. These platforms rely strongly on efficacy estimates based on target-disease association scores created by computational methods for drug target prioritization. Here novel computational methods are presented to more accurately evaluate the efficacy and safety of potential drug targets. The proposed efficacy scores utilize existing gene expression data and tissue/disease specific networks to improve the inference of target-disease associations. Conversely, safety scores enable the identification of genes that are essential, potentially susceptible to adverse effects or carcinogenic. Benchmark results demonstrate that our transcriptome-based methods for drug target prioritization can increase the true positive rate of target-disease associations. Additionally, the proposed safety evaluation system enables accurate predictions of targets of withdrawn drugs and targets of drug trials prematurely discontinued.

Abstract: Identification of community structures in complex network is of crucial importance for understanding the system's function, organization, robustness and security. Here, we present a novel Ollivier-Ricci curvature (ORC) inspired approach to community identification in complex networks. We demonstrate that the intrinsic geometric underpinning of the ORC offers a natural approach to discover inherent community structures within a network based on interaction among entities. We develop an ORC-based community identification algorithm based on the idea of sequential removal of negatively curved edges symptomatic of high interactions (e.g., traffic, attraction). To illustrate and compare the performance with other community identification methods, we examine the ORC-based algorithm with stochastic block model artificial networks and real-world examples ranging from social to drug-drug interaction networks. The ORC-based algorithm is able to identify communities with either better or comparable performance accuracy and to discover finer hierarchical structures of the network. This opens new geometric avenues for analysis of complex networks dynamics.

Abstract: Recent data indicate that up-to 30-40% of cancers can be prevented by dietary and lifestyle measures alone. Herein, we introduce a unique network-based machine learning platform to identify putative food-based cancer-beating molecules. These have been identified through their molecular biological network commonality with clinically approved anti-cancer therapies. A machine-learning algorithm of random walks on graphs (operating within the supercomputing DreamLab platform) was used to simulate drug actions on human interactome networks to obtain genome-wide activity profiles of 1962 approved drugs (199 of which were classified as "anti-cancer" with their primary indications). A supervised approach was employed to predict cancer-beating molecules using these 'learned' interactome activity profiles. The validated model performance predicted anti-cancer therapeutics with classification accuracy of 84-90%. A comprehensive database of 7962 bioactive molecules within foods was fed into the model, which predicted 110 cancer-beating molecules (defined by anti-cancer drug likeness threshold of >70%) with expected capacity comparable to clinically approved anti-cancer drugs from a variety of chemical classes including flavonoids, terpenoids, and polyphenols. This in turn was used to construct a 'food map' with anti-cancer potential of each ingredient defined by the number of cancer-beating molecules found therein. Our analysis underpins the design of next-generation cancer preventative and therapeutic nutrition strategies.

Abstract: Identifying the variants that alter protein function is a promising strategy for deciphering the biological consequences of somatic mutations during tumorigenesis, which could provide novel targets for the development of cancer therapies. Here, based on our previously developed method, we present a strategy called AlloDriver that identifies cancer driver genes/proteins as possible targets from mutations. AlloDriver utilizes structural and dynamic features to prioritize potentially functional genes/proteins in individual cancers via mapping mutations generated from clinical cancer samples to allosteric/orthosteric sites derived from three-dimensional protein structures. This strategy exhibits desirable performance in the reemergence of known cancer driver mutations and genes/proteins from clinical samples. Significantly, the practicability of AlloDriver to discover novel cancer driver proteins in head and neck squamous cell carcinoma (HNSC) was tested in a real case of human protein tyrosine phosphatase, receptor type K (PTPRK) through a L1143F driver mutation located at the allosteric site of PTPRK, which was experimentally validated by cell proliferation assay. AlloDriver is expected to help to uncover innovative molecular mechanisms of tumorigenesis by perturbing proteins and to discover novel targets based on cancer driver mutations. The AlloDriver is freely available to all users at http://mdl.shsmu.edu.cn/ALD.

Abstract: ResponseNet v.3 is an enhanced version of ResponseNet, a web server that is designed to highlight signaling and regulatory pathways connecting user-defined proteins and genes by using the ResponseNet network optimization approach (http://netbio.bgu.ac.il/respnet). Users run ResponseNet by defining source and target sets of proteins, genes and/or microRNAs, and by specifying a molecular interaction network (interactome). The output of ResponseNet is a sparse, high-probability interactome subnetwork that connects the two sets, thereby revealing additional molecules and interactions that are involved in the studied condition. In recent years, massive efforts were invested in profiling the transcriptomes of human tissues, enabling the inference of human tissue interactomes. ResponseNet v.3 expands ResponseNet2.0 by harnessing approximately 11,600 RNA-sequenced human tissue profiles made available by the Genotype-Tissue Expression consortium, to support context-specific analysis of 44 human tissues. Thus, ResponseNet v.3 allows users to illuminate the signaling and regulatory pathways potentially active in the context of a specific tissue, and to compare them with active pathways in other tissues. In the era of precision medicine, such analyses open the door for tissue- and patient-specific analyses of pathways and diseases.

Abstract: Drug combination therapy has the potential to enhance efficacy, reduce dose-dependent toxicity and prevent the emergence of drug resistance. However, discovery of synergistic and effective drug combinations has been a laborious and often serendipitous process. In recent years, identification of combination therapies has been accelerated due to the advances in high-throughput drug screening, but informatics approaches for systems-level data management and analysis are needed. To contribute toward this goal, we created an open-access data portal called DrugComb (https://drugcomb.fimm.fi) where the results of drug combination screening studies are accumulated, standardized and harmonized. Through the data portal, we provided a web server to analyze and visualize users' own drug combination screening data. The users can also effectively participate a crowdsourcing data curation effect by depositing their data at DrugComb. To initiate the data repository, we collected 437 932 drug combinations tested on a variety of cancer cell lines. We showed that linear regression approaches, when considering chemical fingerprints as predictors, have the potential to achieve high accuracy of predicting the sensitivity of drug combinations. All the data and informatics tools are freely available in DrugComb to enable a more efficient utilization of data resources for future drug combination discovery.

Abstract: BACKGROUND: In spite of the abundance of genomic data, predictive models that describe phenotypes as a function of gene expression or mutations are difficult to obtain because they are affected by the curse of dimensionality, given the disbalance between samples and candidate genes. And this is especially dramatic in scenarios in which the availability of samples is difficult, such as the case of rare diseases. RESULTS: The application of multi-output regression machine learning methodologies to predict the potential effect of external proteins over the signaling circuits that trigger Fanconi anemia related cell functionalities, inferred with a mechanistic model, allowed us to detect over 20 potential therapeutic targets. CONCLUSIONS: The use of artificial intelligence methods for the prediction of potentially causal relationships between proteins of interest and cell activities related with disease-related phenotypes opens promising avenues for the systematic search of new targets in rare diseases.