Pathway-based identification of SNPs predictive of survival

Abstract

In recent years, several association analysis methods for case-control studies have been developed. However, as we turn towards the identification of single nucleotide polymorphisms (SNPs) for prognosis, there is a need to develop methods for the identification of SNPs in high dimensional data with survival outcomes. Traditional methods for the identification of SNPs have some drawbacks. First, the majority of the approaches for case-control studies are based on single SNPs. Second, SNPs that are identified without incorporating biological knowledge are more difficult to interpret. Random forests has been found to perform well in gene expression analysis with survival outcomes. In this paper we present the first pathway-based method to correlate SNP with survival outcomes using a machine learning algorithm. We illustrate the application of pathway-based analysis of SNPs predictive of survival with a data set of 192 multiple myeloma patients genotyped for 500 000 SNPs. We also present simulation studies that show that the random forests technique with log-rank score split criterion outperforms several other machine learning algorithms. Thus, pathway-based survival analysis using machine learning tools represents a promising approach for the identification of biologically meaningful SNPs associated with disease.

Random forests for genomic data analysis

“…Pang et al [37, 38] first applied RF on pathway level gene expression data for categorical and continuous phenotypes. RF classification and regression was performed for each pathway using all available samples.…”

“…Pathway testing by RF was extended to censored survival outcomes using random survival forests for both gene expression data and SNP data [40, 38]. An interesting two-stage application of RF pathway analysis was described in Chang et al [41].…”

Pathway analysis of genomic data: concepts, methods, and prospects for future development

“…For example, machine learning approaches [11, 59] attempt to identify the most informative subsets of genes within pathways for association. Networks have been effective in studies of rare variants, as with the identification of a synaptogenesis gene network affected by rare CNVs in autism [60].…”

Big data and computational biology strategy for personalized prognosis

“…It can also provide a robust and synergistic multivariate descriptor of disease complexity via explicit incorporation of the interaction (synergistic) effects of the individual predictors, thereby allowing investigation of a possible classification or prediction model as well as optimizing the predictor subset in input-output interconnections and personalized dose-response relations [10, 44, 45]. Despite known loss of statistical power following dichotomization in the univariate case and in the linear multivariate regression models, it has been shown by many studies that dichotomizing continuous data can greatly improve the power of multiple testing procedures (even in false discovery rate controlling methods) [43].…”

“…Despite known loss of statistical power following dichotomization in the univariate case and in the linear multivariate regression models, it has been shown by many studies that dichotomizing continuous data can greatly improve the power of multiple testing procedures (even in false discovery rate controlling methods) [43]. The appropriate statistical-based predictive models in this case, can lead to unbiased variable selection of highly informative, robust and reproducible components of classifiers and survival predictors [10, 25, 44–46]. It was demonstrated that statistical-based optimization of dichotomous threshold of the continuous variables can be quite accurate, even with highly correlated data [10, 25, 30, 44–47].…”