Comparing machine learning and logistic regression methods for predicting hypertension using a combination of gene expression and next-generation sequencing data

Abstract

Machine learning methods continue to show promise in the analysis of data from genetic association studies because of the high number of variables relative to the number of observations. However, few best practices exist for the application of these methods. We extend a recently proposed supervised machine learning approach for predicting disease risk by genotypes to be able to incorporate gene expression data and rare variants. We then apply 2 different versions of the approach (radial and linear support vector machines) to simulated data from Genetic Analysis Workshop 19 and compare performance to logistic regression. Method performance was not radically different across the 3 methods, although the linear support vector machine tended to show small gains in predictive ability relative to a radial support vector machine and logistic regression. Importantly, as the number of genes in the models was increased, even when those genes contained causal rare variants, model predictive ability showed a statistically significant decrease in performance for both the radial support vector machine and logistic regression. The linear support vector machine showed more robust performance to the inclusion of additional genes. Further work is needed to evaluate machine learning approaches on larger samples and to evaluate the relative improvement in model prediction from the incorporation of gene expression data.

Machine learning and data mining in complex genomic data—a review on the lessons learned in Genetic Analysis Workshop 19

“…In a more direct way, Held et al [ 14 ] built support vector machine models to predict disease status from genes that simultaneously collapse genotype variants and use gene expression effects. Specifically, based on 637 individuals with a simulated hypertension phenotype, some or all of the first 150 simulated data sets were used for a selection of interesting genes (training), and three from the remaining 50 simulated data sets were used for classification (testing).…”

“…The required hyperparameters are derived from cross-validation. Held et al [ 14 ] find that the predictive performance is slightly higher for a support vector machine with a linear kernel than for the other methods. With logistic regression and use of a radial kernel, the performance decreases with a greater number of genes.…”

“…In a most general approach, rare variants are somehow collapsed into some kind of pseudo markers, as seen in the contributions by Sun et al [ 11 ] or Held et al [ 14 ]. This could easily be incorporated in other machine learning approaches as well.…”

“…When focusing on the clinical validity, that is, whether a genetic test is able to predict or identify a disease of interest, both the strength of the association and diagnostic or prognostic value have to be measured. As an example for this, the area under the curve was reported in the contribution by Held et al [ 14 ].…”

An intelligent Alzheimer’s disease diagnosis method using unsupervised feature learning

“…At the second stage, SoftMax regression was applied to classify the health status of individuals using the learned features. Brain images have been obtained under various health conditions [24]. These images have constructed a (6)…”