Screening large-scale association study data: exploiting interactions using random forests

By Kathryn L. Lunetta, L. Hayward, J. Segal, P. van Eerdewegh. 2004

495 citations recorded by the source corpus.

BackgroundGenome-wide association studies for complex diseases will produce genotypes on hundreds of thousands of single nucleotide polymorphisms (SNPs). A logical first approach to dealing with massive numbers of SNPs is to use some test to screen the SNPs, retaining only those that meet some criterion for futher study. For example, SNPs can be ranked by p-value, and those with the lowest p-values retained. When SNPs have large interaction effects but small marginal effects in a population, they are unlikely to be retained when univariate tests are used for screening. However, model-based screens that pre-specify interactions are impractical for data sets with thousands of SNPs. Random forest analysis is an alternative method that produces a single measure of importance for each predictor variable that takes into account interactions among variables without requiring model specification. Interactions increase the importance for the individual interacting variables, making them more likely to be given high importance relative to other variables. We test the performance of random forests as a screening procedure to identify small numbers of risk-associated SNPs from among large numbers of unassociated SNPs using complex disease models with up to 32 loci, incorporating both genetic heterogeneity and multi-locus interaction.ResultsKeeping other factors constant, if risk SNPs interact, the random forest importance measure significantly outperforms the Fisher Exact test as a screening tool. As the number of interacting SNPs increases, the improvement in performance of random forest analysis relative to Fisher Exact test for screening also increases. Random forests perform similarly to the univariate Fisher Exact test as a screening tool when SNPs in the analysis do not interact.ConclusionsIn the context of large-scale genetic association studies where unknown interactions exist among true risk-associated SNPs or SNPs and environmental covariates, screening SNPs using random forest analyses can significantly reduce the number of SNPs that need to be retained for further study compared to standard univariate screening methods.

Subfields: Epidemiology, Genetics, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Epidemiology

In Genetics

In Statistics and Probability

Citations in the corpus

Interpreting Classifiers through Attribute Interactions in Datasets (2017, background): “The beneﬁt of using powerful classiﬁers, such as random forest, is that one does not need to specify the exact form of interactions between attributes (Lunetta et al., 2004), which is necessary in many traditional statistical methods (e.g., linear regression models that include interaction terms).”
A new variable importance measure for random forests with missing data (2012, background): “Moreover, recursive partitioning is able to detect even complex interaction effects, as pointed out by, e.g., Lunetta et al. (2004), Cutler et al. (2007) and Tang et al. (2009).”
Random Forests for Genomic Data Analysis (2012, background): “Lunetta et al. [47] conducted one of the earliest simulation experiments to evaluate the power of RF to screen SNPs with interaction effects in genetic association studies.”
A Bias Correction Algorithm for the Gini Variable Importance Measure in Classification Trees (2008, methodology): “…Gradient Boosting Machine [Friedman(2001)] and similar methods were proposed for solving problems of variable selection and feature extraction in diﬀerent research areas ([Bureau et al. (2003), Guha and Jurs(2004), D´ıaz-Uriarte and Alvarez de Andr´es(2006), Lunetta et al. (2004)], to name a few).”
Overview of random forest methodology and practical guidance with emphasis on computational biology and bioinformatics (2012, background): “They have been considered in tens of bioinformatics papers [81, 62, 80, 2, 44] and biomedical applications [12, 46, 10, 56, 78, 15, 70, 47] including genome-wide studies [63, 79, 31, 64, 51].”
Mining data with random forests: A survey and results of new tests (2011, background): “Dj of RF, Fisher’s Exact test [93] Task: identification of a small number of risk-associated single nucleotide polymorphisms (SNPs) among large number of un-associated SNPs in complex disease models, two data sets of”
Explainable AI for Trees: From Local Explanations to Global Understanding (2019, methodology): “While previous work has used global measures of feature interactions [43, 28], to the best of our knowledge SHAP interaction values represent the first local approach to feature interactions beyond simply listing decision paths.”
Detecting gene-gene interactions using a permutation-based random forest method (2016, methodology): “RF was found to be an successful screening tool which outperformed traditional methods such as Fisher’s exact test for detecting risk associated SNPs by using VIMs when interactions exist [18].”
A Review of Ensemble Methods in Bioinformatics (2010, unknown): “[60] for complex interaction identification.”
Understanding Random Forests: From Theory to Practice (2014, background): “Examples of successful applications include the identification of disease-associated SNPs in genome-wide association studies [Lunetta et al., 2004; Meng et al., 2009; Botta et al., 2014], the discovery of important genes and pathways from micro-array gene-expression data [Pang et al.”
Detecting gene-gene interactions that underlie human diseases (2009, methodology): “( Box 2 ), which has been used in several genetic studie”
An Introduction to Recursive Partitioning: Rationale, Application and Characteristics of Classification and Regression Trees, Bagging and Random Forests (2009, unknown): “For example, Lunetta et al. (2004) found that genetic markers relevant in interactions with other markers or environmental variables can be detected more efficiently by means of random forests than by means of univariate screening methods like Fisher's exact test.”
Data mining in the Life Sciences with Random Forest: a walk in the park or lost in the jungle? (2012, background): “Variable interactions have been reported to increase the global variable importance value [56].”