ExGenet, Integrating Design of Experiments and Response Surface Methodology for Cancer Gene Detection in Gene Regulatory Networks.

Objective: Network analysis techniques often require tuning hyperparameters for optimal performance. For instance, the independent cascade model necessitates determining the probability of diffusion. Despite its importance, a consensus on effective parameter adjustment remains elusive. Methods: In this study, we propose a novel approach utilizing experimental design methodologies, specifically 2-Factorial Analysis for Screening, and Response Surface Methodology (RSM) for parameter adjustment. We apply this methodology to the task of detecting cancer driver genes in colorectal cancer. Result: Through experimental validation of colorectal cancer data, we demonstrate the effectiveness of our proposed methodology. Compared with existing methods, our approach offers several advantages, including reduced computational overhead, systematic parameter selection grounded in statistical theory, and improved performance in detecting cancer driver genes. Conclusion: This study presents a significant advancement in the field of network analysis by providing a practical and systematic approach to hyperparameter tuning. By optimizing parameter settings, our methodology offers promising implications for critical biomedical applications such as cancer driver gene detection.

BuB: a builder-booster model for link prediction on knowledge graphs.

Link prediction (LP) has many applications in various fields. Much research has been carried out on the LP field, and one of the most critical problems in LP models is handling one-to-many and many-to-many relationships. To the best of our knowledge, there is no research on discriminative fine-tuning (DFT). DFT means having different learning rates for every parts of the model. We introduce the BuB model, which has two parts: relationship Builder and Relationship Booster. Relationship Builder is responsible for building the relationship, and Relationship Booster is responsible for strengthening the relationship. By writing the ranking function in polar coordinates and using the nth root, our proposed method provides solutions for handling one-to-many and many-to-many relationships and increases the optimal solutions space. We try to increase the importance of the Builder part by controlling the learning rate using the DFT concept. The experimental results show that the proposed method outperforms state-of-the-art methods on benchmark datasets.

A Social Network Analysis Approach to Evaluate the Relationship Between the Mobility Network Metrics and the COVID-19 Outbreak.

The emergence of the new coronavirus in late 2019 further highlighted the human need for solutions to explore various aspects of deadly pandemics. Providing these solutions will enable humans to be more prepared for dealing with possible future pandemics. In addition, it helps governments implement strategies to tackle and control infectious diseases similar to COVID-19 faster than ever before. In this article, we used the social network analysis (SNA) method to identify high-risk areas of the new coronavirus in Iran. First, we developed the mobility network through the transfer of passengers (edges) between the provinces (nodes) of Iran and then evaluated the in-degree and page rank centralities of the network. Next, we developed 2 Poisson regression (PR) models to predict high-risk areas of the disease in different populations (moderator) using the mobility network centralities (independent variables) and the number of patients (dependent variable). The P-value of .001 for both prediction models confirmed a meaningful interaction between our variables. Besides, the PR models revealed that in higher populations, with the increase of network centralities, the number of patients increases at a higher rate than in lower populations, and vice versa. In conclusion, our method helps governments impose more restrictions on high-risk areas to handle the COVID-19 outbreak and provides a viable solution for accelerating operations against future pandemics similar to the coronavirus.

DGRanker: Cancer Driver Gene Detection in Human Transcriptional Regulatory Network.

Background: Cancer is a group of diseases that have received much attention in biological research because of its high mortality rate and the lack of accurate identification of its root causes. In such studies, researchers usually try to identify cancer driver genes (CDGs) that start cancer in a cell. The majority of the methods that have ever been proposed for the identification of CDGs are based on gene expression data and the concept of mutation in genomic data. Recently, using networking techniques and the concept of influence maximization, some models have been proposed to identify these genes. Objectives: We aimed to construct the cancer transcriptional regulatory network and identify cancer driver genes using a network science approach without the use of mutation and genomic data. Materials and Methods: In this study, we will employ the social influence network theory to identify CDGs in the human gene regulatory network (GRN) that is based on the concept of influence and power of webpages. First, we will create GRN Networks using gene expression data and Existing nodes and edges. Next, we will implement the modified algorithm on GRN networks being studied by weighting the regulatory interaction edges using the influence spread concept. Nodes with the highest ratings will be selected as the CDGs. Results: The results show our proposed method outperforms most of the other computational and network-based methods and show its superiority in identifying CDGs compared to many other methods. In addition, the proposed method can identify many CDGs that are overlooked by all previously published methods. Conclusions: Our study demonstrated that the Google's PageRank algorithm can be utilized and modified as a network-based method for identifying cancer driver gene in transcriptional regulatory network. Furthermore, the proposed method can be considered as a complementary method to the computational-based cancer driver gene identification tools.

GenHITS: A network science approach to driver gene detection in human regulatory network using gene's influence evaluation.

Cancer is among the diseases causing death, in which, cells uncontrollably grow and reproduce beyond the cell regulatory mechanism. In this disease, some genes are initiators of abnormalities and then transmit them to other genes through protein interactions. Accordingly, these genes are known as cancer driver genes (CDGs). In this regard, several methods have been previously developed for identifying cancer driver genes. Most of these methods are computational-based, which use the concept of mutation to predict CDGs. In this research, a method has been proposed for identifying CDGs in the transcription regulatory network using the concept of influence diffusion and by modifying the Hyperlink-Induced Topic Search algorithm based on the diffusion concept. Due to the type of these networks and the processes of abnormality progression in cells and the formation of cancerous tumors, high-influence genes can be the most likely considered as the driver genes. Therefore, we can use the influence diffusion concept as an acceptable theory to identify these genes. Recently, a method has been proposed to detect CDGs with the concept of the influence maximization. One of the challenges in these types of networks is finding the power of regulatory interaction between genes. Moreover, we have proposed a novel method to calculate the weight of regulatory interactions, based on the concept of diffusion. The performance of the proposed method was compared with other seventeen computational and network tools. Correspondingly, three cancer types were used as benchmarks as follows: breast invasive carcinoma (BRCA), Colon adenocarcinoma (COAD), and lung squamous cell carcinoma (LUSC). In addition, to determine the accuracy of the detected drivers using each method, CGC (Cancer Gene Census) and Mut-driver gene lists were utilized as gold standard. The results show that GenHITS performs better compared to the most of the other computational and network methods. Besides, it is also able to identify genes that have been identified by none of the other methods yet.

KatzDriver: A network based method to cancer causal genes discovery in gene regulatory network.

One of the important problems in oncology is finding the genes that perturb the cell functionality and cause cancer. These genes, namely cancer driver genes (CDGs), when mutated, lead to the activation of the abnormal proteins. This abnormality is passed on to other genes by protein-protein interactions, which can cause cells to uncontrollably multiply and become cancerous. So, many methods have been introduced to predict this group of genes. Most of these methods are computational-based, which identify the CDGs based on mutations and genomic data. In this study, we proposed KatzDriver, as a network-based approach, in order to detect CDGs. This method is able to calculate the relative impact of each gene in the spread of abnormality in the gene regulatory network. In this approach, we firstly create the studied networks using gene expression and regulatory interaction data. Then by combining the topological and biological data, the weights of edges (regulatory interactions) and nodes (genes) are calculated. Afterward, based on the KATZ approach, the receiving and broadcasting powers of each gene were calculated to find the relative impact of each gene. At the end, the top genes with the highest relative impact ranks were selected as potential cancer drivers. The result of the proposed approach was compared with 18 existing computational and network-based methods in terms of F-measure, and the number of the predicted cancer driver genes. The result shows that our proposed algorithm is better than most of the other methods. KatzDriver is also able to detect a significant number of unique driver genes compared to other computational and network-based methods.

A new application of community detection for identifying the real specialty of physicians.

BACKGROUND: There is an increasing trend in using network science methods and algorithms, including community detection methods, in different areas of healthcare. These areas include protein networks, drug prescriptions, healthcare fraud detection, and drug abuse. Counterfeit drugs, off-label marketing issues, and finding the healthcare community structures in a network of hospitals, are examples of using community detection in healthcare. OBJECTIVE: This paper attempts to find physicians' real medical specialties based on their prescription history. As a novel application of community detection in the healthcare field, this knowledge can be used as an alternative for missing values of the healthcare databases. Therefore, it could help scientists and researchers to obtain more accurate and more reliable results. METHODS: This research is done through the community detection method and applying big data tools as well as interviews with the field experts. The big data, which is used in this paper, includes 32 million written medical prescriptions in the year 2014, provided by the Health Insurance Organization. The results are validated both qualitatively and quantitatively. RESULTS: The findings reveal nine major communities of physicians, and labeling these communities by experts presents almost every specialty in the drug prescriptions field. Some of these communities are labeled as a single well-known specialty, and some others are consist of two or more specialties that have overlap with each other. CONCLUSION: After receiving the prescription data and getting the experts' opinions, it was revealed that some physicians might persistently prescribe drugs that are not in their fields of expertise. Regarding the accuracy of community detection and the use of existing data values, we proved this hypothesis.

Cancer driver gene discovery in transcriptional regulatory networks using influence maximization approach.

Cancer driver genes (CDGs) are the genes whose mutations cause tumor growth. Several computational methods have been previously developed for finding CDGs. Most of these methods are sequence-based, that is, they rely on finding key mutations in genomic data to predict CDGs. In the present work, we propose iMaxDriver as a network-based tool for predicting driver genes by application of influence maximization algorithm on human transcriptional regulatory network (TRN). In the first step of this approach, the TRN is pruned and weighted by exploiting tumor-specific gene expression (GE) data. Then, influence maximization approach is used to find the influence of each gene. The top genes with the highest influence rate are selected as the potential driver genes. We compared the performance of our CDG prediction method with fifteen other computational tools, based on a benchmark of three different cancer types. Our results show that iMaxDriver outperforms most of the state-of-the-art algorithms for CDG prediction. Furthermore, iMaxDriver is able to correctly predict many CDGs that are overlooked by all previously published tools. Due to this relative orthogonality, iMaxDriver can be considered as a complementary approach to the sequence-based CDG prediction methods.