Feature Selection for Breast Cancer Classification by Integrating Somatic Mutation and Gene Expression

Abstract

Exploring the molecular mechanisms of breast cancer is essential for the early prediction, diagnosis, and treatment of cancer patients. The large scale of data obtained from the high-throughput sequencing technology makes it difficult to identify the driver mutations and a minimal optimal set of genes that are critical to the classification of cancer. In this study, we propose a novel method without any prior information to identify mutated genes associated with breast cancer. For the somatic mutation data, it is processed to a mutated matrix, from which the mutation frequency of each gene can be obtained. By setting a reasonable threshold for the mutation frequency, a mutated gene set is filtered from the mutated matrix. For the gene expression data, it is used to generate the gene expression matrix, while the mutated gene set is mapped onto the matrix to construct a co-expression profile. In the stage of feature selection, we propose a staged feature selection algorithm, using fold change, false discovery rate to select differentially expressed genes, mutual information to remove the irrelevant and redundant features, and the embedded method based on gradient boosting decision tree with Bayesian optimization to obtain an optimal model. In the stage of evaluation, we propose a weighted metric to modify the traditional accuracy to solve the sample imbalance problem. We apply the proposed method to The Cancer Genome Atlas breast cancer data and identify a mutated gene set, among which the implicated genes are oncogenes or tumor suppressors previously reported to be associated with carcinogenesis. As a comparison with the integrative network, we also perform the optimal model on the individual gene expression and the gold standard PMA50. The results show that the integrative network outperforms the gene expression and PMA50 in the average of most metrics, which indicate the effectiveness of our proposed method by integrating multiple data sources, and can discover the associated mutated genes in breast cancer.

Integrating Genomic Information with Tumor-Immune Microenvironment in Triple-Negative Breast Cancer

“…( i) Differences in somatic mutation burden between deceased and alive individuals: The observed differences in somatic mutation profiles between individuals who survived and individuals who died from the disease, and their correlation with tumor stage and metastatic status, suggests that TNBC patients may be amenable to mutation-based stratification. Although mutation-based classification was not attempted in this investigation, owing to the much smaller number of samples in the deceased group, multigene classification and prediction of survival as well as mutation-based classification in breast cancer have been reported [ 42 , 43 , 44 , 45 ], although those reports were not exclusively concerning TNBC. Our investigation examined all somatic mutations, some of which may not be drivers.…”

Familial CCM Genes Might Not Be Main Drivers for Pathogenesis of Sporadic CCMs-Genetic Similarity between Cancers and Vascular Malformations

“…The oncogenic effect of enhanced PI3K signaling in the reproductive system, especially in breast cancers, has been extensively studied [91,[129][130][131]. Cooperation between somatic GOF PIK3CA and LOF p53 mutations have been well documented in breast cancer [19,125,[132][133][134][135][136][137] and reproductive cancers in general [19,82,125,[133][134][135][136][137][138][139][140][141][142].…”

On Combining Feature Selection and Over-Sampling Techniques for Breast Cancer Prediction

“…In [9], a genetic algorithm is used for feature selection, where the selected subset is used to construct different classifiers for performance comparisons. On the other hand, Jiang and Jin [10] use a gradient boosting decision tree with Bayesian optimization to remove the irrelevant and redundant features from gene expression data. Raj et al [11] compare several feature selection methods to determine the best one to combine with the random forest classifier.…”