Jianing Xi

A novel network regularized matrix decomposition method to detect mutated cancer genes in tumour samples with inter-patient heterogeneity

Inter-patient heterogeneity is a major challenge for mutated cancer genes detection which is crucial to advance cancer diagnostics and therapeutics. To detect mutated cancer genes in heterogeneous tumour samples, a prominent strategy is to determine whether the genes are recurrently mutated in their interaction network context. However, recent studies show that some cancer genes in different perturbed pathways are mutated in different subsets of samples. Subsequently, these genes may not display significant mutational recurrence and thus remain undiscovered even in consideration of network information. We develop a novel method called mCGfinder to efficiently detect mutated cancer genes in tumour samples with inter-patient heterogeneity. Based on matrix decomposition framework incorporated with gene interaction network information, mCGfinder can successfully measure the significance of mutational recurrence of genes in a subset of samples. When applying mCGfinder on TCGA somatic mutation datasets of five types of cancers, we find that the genes detected by mCGfinder are significantly enriched for known cancer genes, and yield substantially smaller p-values than other existing methods. All the results demonstrate that mCGfinder is an efficient method in detecting mutated cancer genes.

A novel heterogeneous network-based method for drug response prediction in cancer cell lines

An enduring challenge in personalized medicine lies in selecting a suitable drug for each individual patient. Here we concentrate on predicting drug responses based on a cohort of genomic, chemical structure, and target information. Therefore, a recently study such as GDSC has provided an unprecedented opportunity to infer the potential relationships between cell line and drug. While existing approach rely primarily on regression, classification or multiple kernel learning to predict drug responses. Synthetic approach indicates drug target and protein-protein interaction could have the potential to improve the prediction performance of drug response. In this study, we propose a novel heterogeneous network-based method, named as HNMDRP, to accurately predict cell line-drug associations through incorporating heterogeneity relationship among cell line, drug and target. Compared to previous study, HNMDRP can make good use of above heterogeneous information to predict drug responses. The validity of our method is verified not only by plotting the ROC curve, but also by predicting novel cell line-drug sensitive associations which have dependable literature evidences. This allows us possibly to suggest potential sensitive associations among cell lines and drugs. Matlab and R codes of HNMDRP can be found at following https://github.com/USTC-HIlab/HNMDRP.

DGPathinter: a novel model for identifying driver genes via knowledge-driven matrix factorization with prior knowledge from interactome and pathways

Cataloging mutated driver genes that confer a selective growth advantage for tumor cells from sporadic passenger mutations is a critical problem in cancer genomic research. Previous studies have reported that some driver genes are not highly frequently mutated and cannot be tested as statistically significant, which complicates the identification of driver genes. To address this issue, some existing approaches incorporate prior knowledge from an interactome to detect driver genes which may be dysregulated by interaction network context. However, altered operations of many pathways in cancer progression have been frequently observed, and prior knowledge from pathways is not exploited in the driver gene identification task. In this paper, we introduce a driver gene prioritization method called driver gene identification through pathway and interactome information (DGPathinter), which is based on knowledge-based matrix factorization model with prior knowledge from both interactome and pathways incorporated. When DGPathinter is applied on somatic mutation datasets of three types of cancers and evaluated by known driver genes, the prioritizing performances of DGPathinter are better than the existing interactome driven methods. The top ranked genes detected by DGPathinter are also significantly enriched for known driver genes. Moreover, most of the top ranked scored pathways given by DGPathinter are also cancer progressionassociated pathways. These results suggest that DGPathinter is a useful tool to identify potential driver genes. Subjects Bioinformatics, Computational Biology

Detection of copy number variants and loss of heterozygosity from impure tumor samples using whole exome sequencing data

Using whole-exome sequencing (WES) for the detection of chromosomal aberrations from tumor samples has become increasingly popular, as it is cost-effective and time efficient. However, factors which present in WES tumor samples, including diversity in exon size, batch effect and tumor impurity, can complicate the identification of somatic mutation in each region of the exon. To address these issues, the authors of the present study have developed a novel method, PECNV, for the detection of genomic copy number variants and loss of heterozygosity in WES datasets. PECNV combines normalized logarithm ratio of read counts (Log Ratio) and B allele frequency (BAF), and then employs expectation maximization (EM) algorithm to estimate parameters involved in the models. A comprehensive assessment of PECNV of PECNV was performed by analyzing simulated datasets contaminated with different normal cell proportion and eight real primary triple-negative breast cancer samples. PECNV demonstrated superior results compared with ExomeCNV and EXCAVATOR for the detection of genomic aberrations in WES data.

A novel unsupervised learning model for detecting driver genes from pan-cancer data through matrix tri-factorization framework with pairwise similarities constraints

Abstract Identifying cancer-causing mutated driver genes from passenger mutations is crucial to enhance the development of cancer diagnostics and therapeutics, and many previous efforts have been undertaken to identify cancer driver genes from somatic mutation data of specific types of cancers. However, many driver genes are underestimated when the mutation data of only specific cancers are investigated, which complicates the understanding of tumorigenesis. According to recent studies, cancers of disparate organs have many shared genomic mutations, and some driver genes that are not highly frequently mutated in patients of one cancer type may display considerable mutation frequencies across patients of multiple cancer types. By taking into account both the similarities of mutation profiles of different cancer types and the information of gene interaction network, we propose a novel unsupervised learning model based on matrix tri-factorization by learning the similarities from pairwise constraints to detect driver genes from pan-cancer data. In the evaluation of known benchmarking genes, our model achieves better performance than those of the existing matrix factorization based methods which do not consider the pairwise similarities between cancers. Furthermore, the detection performance of our model is also largely increased (area under the precision-recall curve = 9.1% for Vogelstein genes) when compared with existing methods. Moreover, our model discovers some driver genes that have been reported in recent published studies, showing its potential for application in identifying driver gene candidates for further wet experimental verification.

PTM-ssMP: A Web Server for Predicting Different Types of Post-translational Modification Sites Using Novel Site-specific Modification Profile

Protein post-translational modifications (PTMs) are chemical modifications of a protein after its translation. Owing to its play an important role in deep understanding of various biological processes and the development of effective drugs, PTM site prediction have become a hot topic in bioinformatics. Recently, many online tools are developed to prediction various types of PTM sites, most of which are based on local sequence and some biological information. However, few of existing tools consider the relations between different PTMs for their prediction task. Here, we develop a web server called PTM-ssMP to predict PTM site, which adopts site-specific modification profile (ssMP) to efficiently extract and encode the information of both proximal PTMs and local sequence simultaneously. In PTM-ssMP we provide efficient prediction of multiple types of PTM site including phosphorylation, lysine acetylation, ubiquitination, sumoylation, methylation, O-GalNAc, O-GlcNAc, sulfation and proteolytic cleavage. To assess the performance of PTM-ssMP, a large number of experimentally verified PTM sites are collected from several sources and used to train and test the prediction models. Our results suggest that ssMP consistently contributes to remarkable improvement of prediction performance. In addition, results of independent tests demonstrate that PTM-ssMP compares favorably with other existing tools for different PTM types. PTM-ssMP is implemented as an online web server with user-friendly interface, which is freely available at http://bioinformatics.ustc.edu.cn/PTM-ssMP/index/.

Discovering mutated driver genes through a robust and sparse co-regularized matrix factorization framework with prior information from mRNA expression patterns and interaction network

BackgroundDiscovery of mutated driver genes is one of the primary objective for studying tumorigenesis. To discover some relatively low frequently mutated driver genes from somatic mutation data, many existing methods incorporate interaction network as prior information. However, the prior information of mRNA expression patterns are not exploited by these existing network-based methods, which is also proven to be highly informative of cancer progressions.ResultsTo incorporate prior information from both interaction network and mRNA expressions, we propose a robust and sparse co-regularized nonnegative matrix factorization to discover driver genes from mutation data. Furthermore, our framework also conducts Frobenius norm regularization to overcome overfitting issue. Sparsity-inducing penalty is employed to obtain sparse scores in gene representations, of which the top scored genes are selected as driver candidates. Evaluation experiments by known benchmarking genes indicate that the performance of our method benefits from the two type of prior information. Our method also outperforms the existing network-based methods, and detect some driver genes that are not predicted by the competing methods.ConclusionsIn summary, our proposed method can improve the performance of driver gene discovery by effectively incorporating prior information from interaction network and mRNA expression patterns into a robust and sparse co-regularized matrix factorization framework.