Incorporating Multisource Knowledge To Predict Drug Synergy Based on Graph Co-regularization

A Computational Framework for Predicting Novel Drug Indications Using Graph Convolutional Network With Contrastive Learning

Inferring potential drug indications plays a vital role in the drug discovery process. It can be time-consuming and costly to discover novel drug indications through biological experiments. Recently, graph learning-based methods have gained popularity for this task. These methods typically treat the prediction task as a binary classification problem, focusing on modeling associations between drugs and diseases within a graph. However, labeled data for drug indication prediction is often limited and expensive to acquire. Contrastive learning addresses this challenge by aligning similar drug-disease pairs and separating dissimilar pairs in the embedding space. Thus, we developed a model called DrIGCL for drug indication prediction, which utilizes graph convolutional networks and contrastive learning. DrIGCL incorporates drug structure, disease comorbidities, and known drug indications to extract representations of drugs and diseases. By combining contrastive and classification losses, DrIGCL predicts drug indications effectively. In multiple runs of hold-out validation experiments, DrIGCL consistently outperformed existing computational methods for drug indication prediction, particularly in terms of top-k. Furthermore, our ablation study has demonstrated a significant improvement in the predictive capabilities of our model when utilizing contrastive learning. Finally, we validated the practical usefulness of DrIGCL by examining the predicted novel indications of Aspirin.

An explorative study for leveraging transcriptomic data of embryonic stem cells in mining cancer stemness genes, regulators, and networks

Due to the exquisite ability of cancer stemness to facilitate tumor initiation, metastasis, and cancer therapy resistance, targeting cancer stemness is expected to have clinical implications for cancer treatment. Genes are fundamental for forming and maintaining stemness. Considering shared genetic programs and pathways between embryonic stem cells and cancer stem cells, we conducted a study analyzing transcriptomic data of embryonic stem cells for mining potential cancer stemness genes. Firstly, we integrated co-expression and regression models and predicted 820 stemness genes. Results of gene enrichment analysis confirmed the good prediction performance for enriched signatures in cancer stem cells. Secondly, we provided an application case using the predicted stemness genes to construct a breast cancer stemness network. Mining on the network identified CD44, SOX2, TWIST1, and DLG4 as potential regulators of breast cancer stemness. Thirdly, using the signature of 31,028 chemical perturbations and their correlation with stemness marker genes, we predicted 67 stemness inhibitors with reasonable accuracy of 78%. Two drugs, namely Rigosertib and Proscillaridin A, were first identified as potential stemness inhibitors for melanoma and colon cancer, respectively. Overall, mining embryonic stem cell data provides a valuable way to identify cancer stemness regulators.

Computational models, databases and tools for antibiotic combinations

Antibiotic combination is a promising strategy to extend the lifetime of antibiotics and thereby combat antimicrobial resistance. However, screening for new antibiotic combinations is both time-consuming and labor-intensive. In recent years, an increasing number of researchers have used computational models to predict effective antibiotic combinations. In this review, we summarized existing computational models for antibiotic combinations and discussed the limitations and challenges of these models in detail. In addition, we also collected and summarized available data resources and tools for antibiotic combinations. This study aims to help computational biologists design more accurate and interpretable computational models.

CDCDB: A large and continuously updated drug combination database

In recent years, due to the complementary action of drug combinations over mono-therapy, the multiple-drugs for multiple-targets paradigm has received increased attention to treat bacterial infections and complex diseases. Although new drug combinations screening has benefited from experimental tests like automated high throughput screening, it is limited due to the large number of possible drug combinations. The task of drug combination screening can be streamlined through computational methods and models. Such models require up-to-date databases; however, existing databases are static and consist of the data collected at the time of their creation. This paper introduces the Continuous Drug Combination Database (CDCDB), a continuously updated drug combination database. The CDCDB includes over 40,795 drug combinations, of which 17,107 are unique combinations consisting of more than 4,129 individual drugs, curated from ClinicalTrials.gov, the FDA Orange Book^®, and patents. To create CDCDB, we use various methods, including natural language processing techniques, to improve the process of drug combination discovery, ensuring that our database can be used for drug synergy prediction. Website: https://icc.ise.bgu.ac.il/medical_ai/CDCDB/.

Measurement(s)	drug combination effect modeling • drug combination effect modeling
Technology Type(s)	Text mining • Clinical Trials Informatics System
Factor Type(s)	Medicine
Sample Characteristic - Organism	Homo sapiens

Prediction of Synergistic Antibiotic Combinations by Graph Learning

Antibiotic resistance is a major public health concern. Antibiotic combinations, offering better efficacy at lower doses, are a useful way to handle this problem. However, it is difficult for us to find effective antibiotic combinations in the vast chemical space. Herein, we propose a graph learning framework to predict synergistic antibiotic combinations. In this model, a network proximity method combined with network propagation was used to quantify the relationships of drug pairs, and we found that synergistic antibiotic combinations tend to have smaller network proximity. Therefore, network proximity can be used for building an affinity matrix. Subsequently, the affinity matrix was fed into a graph regularization model to predict potential synergistic antibiotic combinations. Compared with existing methods, our model shows a better performance in the prediction of synergistic antibiotic combinations and interpretability.

Prediction of Synergistic Drug Combinations for Prostate Cancer by Transcriptomic and Network Characteristics

Prostate cancer (PRAD) is a major cause of cancer-related deaths. Current monotherapies show limited efficacy due to often rapidly emerging resistance. Combination therapies could provide an alternative solution to address this problem with enhanced therapeutic effect, reduced cytotoxicity, and delayed the appearance of drug resistance. However, it is prohibitively cost and labor-intensive for the experimental approaches to pick out synergistic combinations from the millions of possibilities. Thus, it is highly desired to explore other efficient strategies to assist experimental researches. Inspired by the challenge, we construct the transcriptomics-based and network-based prediction models to quickly screen the potential drug combination for Prostate cancer, and further assess their performance by in vitro assays. The transcriptomics-based method screens nine possible combinations. However, the network-based method gives discrepancies for at least three drug pairs. Further experimental results indicate the dose-dependent effects of the three docetaxel-containing combinations, and confirm the synergistic effects of the other six combinations predicted by the transcriptomics-based model. For the network-based predictions, in vitro tests give opposite results to the two combinations (i.e. mitoxantrone-cyproheptadine and cabazitaxel-cyproheptadine). Namely, the transcriptomics-based method outperforms the network-based one for the specific disease like Prostate cancer, which provide guideline for selection of the computational methods in the drug combination screening. More importantly, six combinations (the three mitoxantrone-containing and the three cabazitaxel-containing combinations) are found to be promising candidates to synergistically conquer Prostate cancer.

An In Silico Method for Predicting Drug Synergy Based on Multitask Learning

To make better use of all kinds of knowledge to predict drug synergy, it is crucial to successfully establish a drug synergy prediction model and leverage the reconstruction of sparse known drug targets. Therefore, we present an in silico method that predicts the synergy scores of drug pairs based on multitask learning (DSML) that could fuse drug targets, protein-protein interactions, anatomical therapeutic chemical codes, a priori knowledge of drug combinations. To simultaneously reconstruct drug-target protein interactions and synergistic drug combinations, DSML benefits indirectly from the associations with relation through proteins. In cross-validation experiments, DSML improved the ability to predict drug synergy. Moreover, the reconstruction of drug-target interactions and the incorporation of multisource knowledge significantly improved drug combination predictions by a large margin. The potential drug combinations predicted by DSML demonstrate its ability to predict drug synergy.

IDDkin: Network-based influence deep diffusion model for enhancing prediction of kinase inhibitors

MOTIVATION

Protein kinases have been the focus of drug discovery research for many years because they play a causal role in many human diseases. Understanding the binding profile of kinase inhibitors is a prerequisite for drug discovery, and traditional methods of predicting kinase inhibitors are time-consuming and inefficient. Calculation-based predictive methods provide a relatively low-cost and high-efficiency approach to the rapid development and effective understanding of the binding profile of kinase inhibitors. Particularly, the continuous improvement of network pharmacology methods provides unprecedented opportunities for drug discovery, network-based computational methods could be employed to aggregate the effective information from heterogeneous sources, which have become a new way for predicting the binding profile of kinase inhibitors.

RESULTS

In this study, we proposed a network-based influence deep diffusion model, named IDDkin, for enhancing the prediction of kinase inhibitors. IDDkin uses deep graph convolutional networks, graph attention networks and adaptive weighting methods to diffuse the effective information of heterogeneous networks. The updated kinase and compound representations are used to predict potential compound-kinase pairs. The experimental results show that the performance of IDDkin is superior to the comparison methods, including the state-of-the art kinase inhibitor prediction method and the classic model widely used in relationship prediction. In experiments conducted to verify its generalizability and in case studies, the IDDkin model also shows excellent performance. All of these results demonstrate the powerful predictive ability of the IDDkin model in the field of kinase inhibitors.

AVAILABILITY AND IMPLEMENTATION

Source code and data can be downloaded from https://github.com/ CS-BIO/IDDkin.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.