Finding Important Genes from High-Dimensional Data: An Appraisal of Statistical Tests and Machine-Learning Approaches
This work addresses the problem of unreliable gene selection in high-dimensional data analysis for researchers in bioinformatics and statistics, highlighting incremental improvements in model evaluation and selection.
The authors investigated the limitations and misuses of statistical and machine-learning tools for gene selection from high-dimensional cancer data, finding that models achieving 100% accuracy often select inconsistent gene sets and can classify data without using disease-relevant variables. They demonstrated that stochastic gradient boosting is a superior model for gene selection and variable screening under moderate sample sizes and suitable pre-screening.
Over the past decades, statisticians and machine-learning researchers have developed literally thousands of new tools for the reduction of high-dimensional data in order to identify the variables most responsible for a particular trait. These tools have applications in a plethora of settings, including data analysis in the fields of business, education, forensics, and biology (such as microarray, proteomics, brain imaging), to name a few. In the present work, we focus our investigation on the limitations and potential misuses of certain tools in the analysis of the benchmark colon cancer data (2,000 variables; Alon et al., 1999) and the prostate cancer data (6,033 variables; Efron, 2010, 2008). Our analysis demonstrates that models that produce 100% accuracy measures often select different sets of genes and cannot stand the scrutiny of parameter estimates and model stability. Furthermore, we created a host of simulation datasets and "artificial diseases" to evaluate the reliability of commonly used statistical and data mining tools. We found that certain widely used models can classify the data with 100% accuracy without using any of the variables responsible for the disease. With moderate sample size and suitable pre-screening, stochastic gradient boosting will be shown to be a superior model for gene selection and variable screening from high-dimensional datasets.