Network-based Biased Tree Ensembles (NetBiTE) for Drug Sensitivity Prediction and Drug Sensitivity Biomarker Identification in Cancer

Abstract We present the Network-based Biased Tree Ensembles (NetBiTE) method for drug sensitivity prediction and drug sensitivity biomarker identification in cancer using a combination of prior knowledge and gene expression data. Our devised method consists of a biased tree ensemble that is built according to a probabilistic bias weight distribution. The bias weight distribution is obtained from the assignment of high weights to the drug targets and propagating the assigned weights over a protein-protein interaction network such as STRING. The propagation of weights, defines neighborhoods of influence around the drug targets and as such simulates the spread of perturbations within the cell, following drug administration. Using a synthetic dataset, we showcase how application of biased tree ensembles (BiTE) results in significant accuracy gains at a much lower computational cost compared to the unbiased random forests (RF) algorithm. We then apply NetBiTE to the Genomics of Drug Sensitivity in Cancer (GDSC) dataset and demonstrate that NetBiTE outperforms RF in predicting IC50 drug sensitivity, only for drugs that target membrane receptor pathways (MRPs): RTK, EGFR and IGFR signaling pathways. We propose based on the NetBiTE results, that for drugs that inhibit MRPs, the expression of target genes prior to drug administration is a biomarker for IC50 drug sensitivity following drug administration. We further verify and reinforce this proposition through control studies on, PI3K/MTOR signaling pathway inhibitors, a drug category that does not target MRPs, and through assignment of dummy targets to MRP inhibiting drugs and investigating the variation in NetBiTE accuracy.

Download Full-text

Feature selection strategies for drug sensitivity prediction

10.1101/856013 ◽

2019 ◽

Author(s):

Krzysztof Koras ◽

Dilafruz Juraeva ◽

Julian Kreis ◽

Johanna Mazur ◽

Eike Staub ◽

...

Keyword(s):

Gene Expression ◽

Feature Selection ◽

Prior Knowledge ◽

Drug Targets ◽

Drug Response ◽

Drug Sensitivity ◽

Selection Strategies ◽

Genome Wide ◽

Gene Expression Signatures ◽

Drug Sensitivity Prediction

Drug sensitivity prediction constitutes one of the main challenges in personalized medicine. The major difficulty of this problem stems from the fact that the sensitivity of cancer cells to treatment depends on an unknown subset of a large number of biological features. Although feature selection is the key to interpretable results and identification of potential biomarkers, a comprehensive assessment of feature selection methods for drug sensitivity prediction has so far not been performed. We propose feature selection approaches driven by prior knowledge of drug targets, target pathways, and gene expression signatures. We asses these methodologies on Genomics of Drug Sensitivity in Cancer (GDSC) dataset, a panel of around 1000 cell lines screened against multiple anticancer compounds. We compare our results with a baseline model utilizing genome-wide gene expression features and common data-driven feature selection techniques. Together, 2484 unique models were evaluated, providing a comprehensive study of feature selection strategies for the drug response prediction problem. For 23 drugs, the models achieve better predictive performance when the features are selected according to prior knowledge of drug targets and pathways. The best correlation of observed and predicted response using the test set is achieved for Linifanib (r=0.75). Extending the drug-dependent features with gene expression signatures yields models that are most predictive of drug response for 60 drugs, with the best performing example of Dabrafenib. Examples of how pre-selection of features benefits the model interpretability are given for Dabrafenib, Linifanib and Quizartinib. Based on GDSC drug data, we find that feature selection driven by prior knowledge tends to yield better results for drugs targeting specific genes and pathways, while models with the genome-wide features perform better for drugs affecting general mechanisms such as metabolism and DNA replication. For a significant group of the compounds, even a very small number of features based on simple drug properties is often highly predictive of drug sensitivity, can explain the mechanism of drug action and be used as guidelines for their prescription. In general, choosing appropriate feature selection strategies has the potential to develop interpretable models that are indicative for therapy design.

Download Full-text

Discussion on the Mechanism of Lithospermum erythrorhizon in the Treatment of Cervical Cancer Based on Network Pharmacology and Molecular Docking Technology

Tobacco Regulatory Science ◽

10.18001/trs.7.5.1.165 ◽

2021 ◽

Vol 7 (5) ◽

pp. 3927-3933

Author(s):

Qiong Yan ◽

Fangwu Ye

Keyword(s):

Cervical Cancer ◽

Molecular Docking ◽

Drug Targets ◽

Target Genes ◽

Interaction Network ◽

Network Pharmacology ◽

Drug Candidate ◽

Pathway Enrichment Analysis ◽

Active Ingredients ◽

Stable Conformation

Objective To explore the “multi-component, multi-target, multi-pathway” mechanism of Lithospermum erythrorhizonagairtst cervical cancer. Methods The active ingredients and corresponding targets were screened through TCMSP, PubChem and SwissTargetPrediction databases. The GeneCarts platform was used to collect cervical cancer-related genes, and the intersection of drug targets and cervical cancer targets was analyzed. Use STRING to analyze protein interaction network, use Cytoscape software to construct component-target and core target interaction network, perform KEGG pathway enrichment analysis on core target genes, and conduct molecular docking verification.Results After screening, 12 main active ingredients of comfrey (including Shikonin A, 1-methoxyacetylshikonin, Shikonin B, etc.) and 35 key targets related to comfrey and cervical cancer were obtained (including ESR1, SRC, MMP9, PTGS2, etc.). And these genes were mainly enriched in 39 signaling pathways such as PI3K-Akt and estrogen. Molecular docking reminder that Lithospermum A has a higher affinity with ESR1, and Lithospermum B can form a stable conformation with SRC, MMP9, and PTGS2. Conclusion Lithospermum erythrorhizon is a potential drug candidate for the treatment of cervical cancer. It can treat cervical cancer through multi-component, multi-target, and multi-channel action.

Download Full-text

Identifying Alzheimer’s Disease-related miRNA Based on Semi-clustering

Current Gene Therapy ◽

10.2174/1566523219666190924113737 ◽

2019 ◽

Vol 19 (4) ◽

pp. 216-223 ◽

Cited By ~ 2

Author(s):

Tianyi Zhao ◽

Donghua Wang ◽

Yang Hu ◽

Ningyi Zhang ◽

Tianyi Zang ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Drug Targets ◽

Molecular Mechanisms ◽

Feature Vector ◽

Mirna Gene ◽

Interaction Network ◽

Gene Interaction ◽

Proteinprotein Interaction ◽

Synaptic Structures

Background: More and more scholars are trying to use it as a specific biomarker for Alzheimer’s Disease (AD) and mild cognitive impairment (MCI). Multiple studies have indicated that miRNAs are associated with poor axonal growth and loss of synaptic structures, both of which are early events in AD. The overall loss of miRNA may be associated with aging, increasing the incidence of AD, and may also be involved in the disease through some specific molecular mechanisms. Objective: Identifying Alzheimer’s disease-related miRNA can help us find new drug targets, early diagnosis. Materials and Methods: We used genes as a bridge to connect AD and miRNAs. Firstly, proteinprotein interaction network is used to find more AD-related genes by known AD-related genes. Then, each miRNA’s correlation with these genes is obtained by miRNA-gene interaction. Finally, each miRNA could get a feature vector representing its correlation with AD. Unlike other studies, we do not generate negative samples randomly with using classification method to identify AD-related miRNAs. Here we use a semi-clustering method ‘one-class SVM’. AD-related miRNAs are considered as outliers and our aim is to identify the miRNAs that are similar to known AD-related miRNAs (outliers). Results and Conclusion: We identified 257 novel AD-related miRNAs and compare our method with SVM which is applied by generating negative samples. The AUC of our method is much higher than SVM and we did case studies to prove that our results are reliable.

Download Full-text

Bioinformatics Analysis Reveals Functions of MicroRNAs in Rice Under the Drought Stress

Current Bioinformatics ◽

10.2174/1574893615666200207092410 ◽

2021 ◽

Vol 15 (8) ◽

pp. 927-936 ◽

Cited By ~ 3

Author(s):

Yan Peng ◽

Yuewu Liu ◽

Xinbo Chen

Keyword(s):

Drought Stress ◽

Target Genes ◽

Hot Spot ◽

Interaction Network ◽

Enrichment Analysis ◽

Differentially Expressed ◽

Protein Protein Interaction ◽

Promising Tool ◽

Differentially Expressed Mirnas ◽

Aba Biosynthesis

Background: Drought is one of the most damaging and widespread abiotic stresses that can severely limit the rice production. MicroRNAs (miRNAs) act as a promising tool for improving the drought tolerance of rice and have become a hot spot in recent years. Objective: In order to further extend the understanding of miRNAs, the functions of miRNAs in rice under drought stress are analyzed by bioinformatics. Method: In this study, we integrated miRNAs and genes transcriptome data of rice under the drought stress. Some bioinformatics methods were used to reveal the functions of miRNAs in rice under drought stress. These methods included target genes identification, differentially expressed miRNAs screening, enrichment analysis of DEGs, network constructions for miRNA-target and target-target proteins interaction. Results: (1) A total of 229 miRNAs with differential expression in rice under the drought stress, corresponding to 73 rice miRNAs families, were identiﬁed. (2) 1035 differentially expressed genes (DEGs) were identified, which included 357 up-regulated genes, 542 down-regulated genes and 136 up/down-regulated genes. (3) The network of regulatory relationships between 73 rice miRNAs families and 1035 DEGs was constructed. (4) 25 UP_KEYWORDS terms of DEGs, 125 GO terms and 7 pathways were obtained. (5) The protein-protein interaction network of 1035 DEGs was constructed. Conclusion: (1) MiRNA-regulated targets in rice might mainly involve in a series of basic biological processes and pathways under drought conditions. (2) MiRNAs in rice might play critical roles in Lignin degradation and ABA biosynthesis. (3) MiRNAs in rice might play an important role in drought signal perceiving and transduction.

Download Full-text

Cellular Fitness Phenotypes of Cancer Target Genes from Oncobiology to Cancer Therapeutics

Cells ◽

10.3390/cells10020433 ◽

2021 ◽

Vol 10 (2) ◽

pp. 433

Author(s):

Bijesh George ◽

P. Mukundan Pillai ◽

Aswathy Mary Paul ◽

Revikumar Amjesh ◽

Kim Leitzel ◽

...

Keyword(s):

Drug Targets ◽

Target Genes ◽

Human Cancer ◽

Cancer Therapeutics ◽

Survival Duration ◽

Cancer Drug ◽

Targeted Therapeutics ◽

Cellular Targets ◽

Human Cancer Cells ◽

Cancer Types

To define the growing significance of cellular targets and/or effectors of cancer drugs, we examined the fitness dependency of cellular targets and effectors of cancer drug targets across human cancer cells from 19 cancer types. We observed that the deletion of 35 out of 47 cellular effectors and/or targets of oncology drugs did not result in the expected loss of cell fitness in appropriate cancer types for which drugs targeting or utilizing these molecules for their actions were approved. Additionally, our analysis recognized 43 cellular molecules as fitness genes in several cancer types in which these drugs were not approved, and thus, providing clues for repurposing certain approved oncology drugs in such cancer types. For example, we found a widespread upregulation and fitness dependency of several components of the mevalonate and purine biosynthesis pathways (currently targeted by bisphosphonates, statins, and pemetrexed in certain cancers) and an association between the overexpression of these molecules and reduction in the overall survival duration of patients with breast and other hard-to-treat cancers, for which such drugs are not approved. In brief, the present analysis raised cautions about off-target and undesirable effects of certain oncology drugs in a subset of cancers where the intended cellular effectors of drug might not be good fitness genes and that this study offers a potential rationale for repurposing certain approved oncology drugs for targeted therapeutics in additional cancer types.

Download Full-text

Deciphering the signaling network of breast cancer improves drug sensitivity prediction

Cell Systems ◽

10.1016/j.cels.2021.04.002 ◽

2021 ◽

Author(s):

Marco Tognetti ◽

Attila Gabor ◽

Mi Yang ◽

Valentina Cappelletti ◽

Jonas Windhager ◽

...

Keyword(s):

Breast Cancer ◽

Drug Sensitivity ◽

Signaling Network ◽

Drug Sensitivity Prediction

Download Full-text

Identifying Personalized Metabolic Signatures in Breast Cancer

Metabolites ◽

10.3390/metabo11010020 ◽

2020 ◽

Vol 11 (1) ◽

pp. 20

Author(s):

Priyanka Baloni ◽

Wikum Dinalankara ◽

John C. Earls ◽

Theo A. Knijnenburg ◽

Donald Geman ◽

...

Keyword(s):

Drug Targets ◽

Metabolic Networks ◽

Target Genes ◽

Enrichment Analysis ◽

Metabolic Reprogramming ◽

The Cancer Genome Atlas ◽

Estrogen Metabolism ◽

Acid Oxidation ◽

A Genome ◽

Context Specific

Cancer cells are adept at reprogramming energy metabolism, and the precise manifestation of this metabolic reprogramming exhibits heterogeneity across individuals (and from cell to cell). In this study, we analyzed the metabolic differences between interpersonal heterogeneous cancer phenotypes. We used divergence analysis on gene expression data of 1156 breast normal and tumor samples from The Cancer Genome Atlas (TCGA) and integrated this information with a genome-scale reconstruction of human metabolism to generate personalized, context-specific metabolic networks. Using this approach, we classified the samples into four distinct groups based on their metabolic profiles. Enrichment analysis of the subsystems indicated that amino acid metabolism, fatty acid oxidation, citric acid cycle, androgen and estrogen metabolism, and reactive oxygen species (ROS) detoxification distinguished these four groups. Additionally, we developed a workflow to identify potential drugs that can selectively target genes associated with the reactions of interest. MG-132 (a proteasome inhibitor) and OSU-03012 (a celecoxib derivative) were the top-ranking drugs identified from our analysis and known to have anti-tumor activity. Our approach has the potential to provide mechanistic insights into cancer-specific metabolic dependencies, ultimately enabling the identification of potential drug targets for each patient independently, contributing to a rational personalized medicine approach.

Download Full-text

A Multi-Pronged Computational Pipeline for Prioritizing Drug Target Strategies for Latent Tuberculosis

Frontiers in Chemistry ◽

10.3389/fchem.2020.593497 ◽

2020 ◽

Vol 8 ◽

Author(s):

Ushashi Banerjee ◽

Santhosh Sankar ◽

Amit Singh ◽

Nagasuma Chandra

Keyword(s):

Drug Targets ◽

Metabolic Flux ◽

Latent Tuberculosis ◽

Interaction Network ◽

New Drugs ◽

Prolonged Treatment ◽

Flux Analysis ◽

Protein Protein Interaction ◽

Candidate Drug ◽

Global Population

Tuberculosis is one of the deadliest infectious diseases worldwide and the prevalence of latent tuberculosis acts as a huge roadblock in the global effort to eradicate tuberculosis. Most of the currently available anti-tubercular drugs act against the actively replicating form of Mycobacterium tuberculosis (Mtb), and are not effective against the non-replicating dormant form present in latent tuberculosis. With about 30% of the global population harboring latent tuberculosis and the requirement for prolonged treatment duration with the available drugs in such cases, the rate of adherence and successful completion of therapy is low. This necessitates the discovery of new drugs effective against latent tuberculosis. In this work, we have employed a combination of bioinformatics and chemoinformatics approaches to identify potential targets and lead candidates against latent tuberculosis. Our pipeline adopts transcriptome-integrated metabolic flux analysis combined with an analysis of a transcriptome-integrated protein-protein interaction network to identify perturbations in dormant Mtb which leads to a shortlist of 6 potential drug targets. We perform a further selection of the candidate targets and identify potential leads for 3 targets using a range of bioinformatics methods including structural modeling, binding site association and ligand fingerprint similarities. Put together, we identify potential new strategies for targeting latent tuberculosis, new candidate drug targets as well as important lead clues for drug design.

Download Full-text

Decoding the molecular mechanism of parthenocarpy in Musa spp. through protein–protein interaction network

Scientific Reports ◽

10.1038/s41598-021-93661-3 ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Suthanthiram Backiyarani ◽

Rajendran Sasikala ◽

Simeon Sharmiladevi ◽

Subbaraya Uma

Keyword(s):

Candidate Genes ◽

Protein Interaction ◽

Target Genes ◽

Interaction Network ◽

Enrichment Analysis ◽

Auxin Signaling ◽

Pathway Enrichment Analysis ◽

Ppi Network ◽

Protein Protein Interaction ◽

The Difference

AbstractBanana, one of the most important staple fruit among global consumers is highly sterile owing to natural parthenocarpy. Identification of genetic factors responsible for parthenocarpy would facilitate the conventional breeders to improve the seeded accessions. We have constructed Protein–protein interaction (PPI) network through mining differentially expressed genes and the genes used for transgenic studies with respect to parthenocarpy. Based on the topological and pathway enrichment analysis of proteins in PPI network, 12 candidate genes were shortlisted. By further validating these candidate genes in seeded and seedless accession of Musa spp. we put forward MaAGL8, MaMADS16, MaGH3.8, MaMADS29, MaRGA1, MaEXPA1, MaGID1C, MaHK2 and MaBAM1 as possible target genes in the study of natural parthenocarpy. In contrary, expression profile of MaACLB-2 and MaZEP is anticipated to highlight the difference in artificially induced and natural parthenocarpy. By exploring the PPI of validated genes from the network, we postulated a putative pathway that bring insights into the significance of cytokinin mediated CLAVATA(CLV)–WUSHEL(WUS) signaling pathway in addition to gibberellin mediated auxin signaling in parthenocarpy. Our analysis is the first attempt to identify candidate genes and to hypothesize a putative mechanism that bridges the gaps in understanding natural parthenocarpy through PPI network.

Download Full-text