Search

Omics data usually have ultra-high dimension (p) and small sample size (n). Standard support vector machines (SVMs), which minimize the L 2 norm for the primal variables, only lead to sparse solutions for the dual variables. L 1 based SVMs, directly minimizing the L 1 norm, have been used for feature selection with omics data. However, most current methods directly solve the primal formulations of the problem, which are not computationally scalable. The computational complexity increases with the number of features. In addition, L 1 norm is known to be asymptotically biased and not consistent for feature selection. In this paper, we develop an efficient method for sparse support vector machines with L 0 norm approximation. The proposed method approximates the L 0 minimization through solving a series of L 2 optimization problems, which can be formulated with dual variables. It finds the optimal solution for p primal variables through estimating n dual variables, which is more efficient as long as the sample size is small. L 0 approximation leads to sparsity in both dual and primal variables, and can be used for both feature and sample selections. The proposed method identifies much less number of features and achieves similar performances in simulations. We apply the proposed method to feature selections with metagenomic sequencing and gene expression data. It can identify biologically important genes and taxa efficiently., (Copyright © 2019 Elsevier B.V. All rights reserved.)

ObjectiveSuboptimal cytoreductive surgery in advanced epithelial ovarian cancer (EOC) is associated with poor survival but it is unknown if poor outcome is due to the intrinsic biology of unresectable tumors or insufficient surgical effort resulting in residual tumor-sustaining clones. Our objective was to identify the potential molecular pathway(s) and cell type(s) that may be responsible for suboptimal surgical resection.MethodsBy comparing gene expression in optimally and suboptimally cytoreduced patients, we identified a gene network associated with suboptimal cytoreduction and explored the biological processes and cell types associated with this gene network.ResultsWe show that primary tumors from suboptimally cytoreduced patients express molecular signatures that are typically present in a distinct molecular subtype of EOC characterized by increased stromal activation and lymphovascular invasion. Similar molecular pathways are present in EOC metastases, suggesting that primary tumors in suboptimally cytoreduced patients are biologically similar to metastatic tumors. We demonstrate that the suboptimal cytoreduction network genes are enriched in reactive tumor stroma cells rather than malignant tumor cells.ConclusionOur data suggest that the success of cytoreductive surgery is dictated by tumor biology, such as extensive stromal reaction and increased invasiveness, which may hinder surgical resection and ultimately lead to poor survival.

Development of an integrated time and dose model to explore the dynamics of gene expression alterations and identify biomarkers for biodosimetry following low- and high-dose irradiations at high dose rate. We utilized multiple transcriptome datasets (GSE8917, GSE43151, and GSE23515) from Gene Expression Omnibus (GEO) for identifying candidate biological dosimeters. A linear mixed-effects model with random intercept was used to explore the dose-time dynamics of transcriptional responses and to functionally characterize the time- and dose-dependent changes in gene expression. We identified genes that are correlated with dose and time and discovered two clusters of genes that are either positively or negatively correlated with both dose and time based on the parameters of the model. Genes in these two clusters may have persistent transcriptional alterations. Twelve potential transcriptional markers for dosimetry—ARHGEF3, BAX, BBC3, CCDC109B, DCP1B, DDB2, F11R, GADD45A, GSS, PLK3, TNFRSF10B, and XPC were identified. Of these genes, BAX, GSS, and TNFRSF10B are positively associated with both dose and time course, have a persistent transcriptional response, and might be better biological dosimeters. With the proposed approach, we may identify candidate biomarkers that change monotonically in relation to dose, have a persistent transcriptional response, and are reliable over a wide dose range. [ABSTRACT FROM AUTHOR]

Triple-negative breast cancer (TNBC) is generally considered an aggressive breast cancer subtype associated with poor prognostic outcomes. Up to now, the molecular and cellular mechanisms underlying TNBC pathology have not been fully understood. In this manuscript, we propose a novel semiparametric model with kernel for gene-based analysis with a breast cancer GWAS data. The software of SPMGBA (semiparametric method for gene-based analysis) in MATLAB is available at GitHub (https://github.com/zliu3/SPMGBA). Genetic signatures associated with breast cancer are discovered. We further validate the prognostic power of the identified genes with a large cohort of expression data from the European Genome-Phenome Archive, and discover that SEL1L is associated with the overall survival of TNBC with the p-value of.0002. We conclude that gene SEL1L is down-regulated in TNBC and the expression of SEL1L is positively associated with patient survival. [ABSTRACT FROM AUTHOR]

Omics data usually have ultra-high dimension (p) and small sample size (n). Standard support vector machines (SVMs), which minimize the L2 norm for the primal variables, only lead to sparse solutions for the dual variables. L1 based SVMs, directly minimizing the L1 norm, have been used for feature selection with omics data. However, most current methods directly solve the primal formulations of the problem, which are not computationally scalable. The computational complexity increases with the number of features. In addition, L1 norm is known to be asymptotically biased and not consistent for feature selection. In this paper, we develop an efficient method for sparse support vector machines with L0 norm approximation. The proposed method approximates the L0 minimization through solving a series of L2 optimization problems, which can be formulated with dual variables. It finds the optimal solution for p primal variables through estimating n dual variables, which is more efficient as long as the sample size is small. L0 approximation leads to sparsity in both dual and primal variables, and can be used for both feature and sample selections. The proposed method identifies much less number of features and achieves similar performances in simulations. We apply the proposed method to feature selections with metagenomic sequencing and gene expression data. It can identify biologically important genes and taxa efficiently. [ABSTRACT FROM AUTHOR]

Although cancer-associated fibroblasts (CAFs) are viewed as a promising therapeutic target, the design of rational therapy has been hampered by two key obstacles. First, attempts to ablate CAFs have resulted in significant toxicity because currently used biomarkers cannot effectively distinguish activated CAFs from non-cancer associated fibroblasts and mesenchymal progenitor cells. Second, it is unclear whether CAFs in different organs have different molecular and functional properties that necessitate organ-specific therapeutic designs. Our analyses uncovered COL11A1 as a highly specific biomarker of activated CAFs. Using COL11A1 as a 'seed', we identified co-expressed genes in 13 types of primary carcinoma in The Cancer Genome Atlas. We demonstrated that a molecular signature of activated CAFs is conserved in epithelial cancers regardless of organ site and transforming events within cancer cells, suggesting that targeting fibroblast activation should be effective in multiple cancers. We prioritized several potential pan-cancer therapeutic targets that are likely to have high specificity for activated CAFs and minimal toxicity in normal tissues. [ABSTRACT FROM AUTHOR]

Kernel principal component analysis (KPCA) has been applied to data clustering and graphic cut in the last couple of years. This paper discusses the application of KPCA to microarray data clustering. A new algorithm based on KPCA and fuzzy C-means is proposed. Experiments with microarray data show that the proposed algorithms is in general superior to traditional algorithms. [ABSTRACT FROM AUTHOR]

Neuroblastoma is a common cancer in children, affected by a number of genes that interact with each other through intricate but coordinated networks. Traditional approaches can only reconstruct a single regulatory network that is topologically not informative enough to explain the complexity of neuroblastoma risk. We implemented and modified an advanced model for recovering informative, omnidirectional, dynamic, and personalized networks (idopNetworks) from static gene expression data for neuroblastoma risk. We analyzed 3439 immune genes of neuroblastoma for 217 high-risk patients and 30 low-risk patients by which to reconstruct large patient-specific idopNetworks. By converting these networks into risk-specific representations, we found that the shift in patients from a low to high risk or from a high to low risk might be due to the reciprocal change of hub regulators. By altering the directions of regulation exerted by these hubs, it may be possible to reduce a high risk to a low risk. Results from a holistic, systems-oriented paradigm through idopNetworks can potentially enable oncologists to experimentally identify the biomarkers of neuroblastoma and other cancers. [ABSTRACT FROM AUTHOR]

Abstract: Analysis of molecular genetic markers in biological fluids has been proposed as a useful tool for cancer diagnosis. MicroRNAs (miRNAs) are small regulatory RNAs that are frequently dysregulated in lung cancer and have shown promise as tissue-based markers for its prognostication. The aim of this study was to determine whether aberrant miRNA expression can be used as a marker in sputum specimen for the diagnosis of non-small cell lung cancer (NSCLC). Experimental design: expressions of mature miRNAs, mir-21 and mir-155, were examined by real-time reverse transcription polymerase chain reaction (RT-PCR) and normalized to that of control miRNA, U6B, in sputum of 23 patients with NSCLC and 17 cancer-free subjects. The data was compared with conventional sputum cytology for the diagnosis of lung cancer. All endogenous miRNAs were present in sputum in a remarkably stable form and sensitively and specifically detected by real-time RT-PCR. Mir-21 expression in the sputum specimens was significantly higher in cancer patients (76.32±9.79) than cancer-free individuals (62.24±3.82) (P <0.0001). Furthermore, overexpression of mir-21 showed highly discriminative receiver-operator characteristic (ROC) curve profile, clearly distinguishing cancer patients from cancer-free subjects with areas under the ROC curve at 0.902±0.054. Detection of mir-21 expression produced 69.66% sensitivity and 100.00% specificity in diagnosis of lung cancer, as compared with 47.82% sensitivity and 100.00% specificity by sputum cytology. The measurement of altered miRNA expression in sputum could be a useful noninvasive approach for the diagnosis of lung cancer. [Copyright &y& Elsevier]