Role of genetic heterogeneity and epistasis in bladder cancer susceptibility and outcome: a learning classifier system approach

Abstract

Background and objectiveDetecting complex patterns of association between genetic or environmental risk factors and disease risk has become an important target for epidemiological research. In particular, strategies that provide multifactor interactions or heterogeneous patterns of association can offer new insights into association studies for which traditional analytic tools have had limited success.Materials and methodsTo concurrently examine these phenomena, previous work has successfully considered the application of learning classifier systems (LCSs), a flexible class of evolutionary algorithms that distributes learned associations over a population of rules. Subsequent work dealt with the inherent problems of knowledge discovery and interpretation within these algorithms, allowing for the characterization of heterogeneous patterns of association. Whereas these previous advancements were evaluated using complex simulation studies, this study applied these collective works to a ‘real-world’ genetic epidemiology study of bladder cancer susceptibility.Results and discussionWe replicated the identification of previously characterized factors that modify bladder cancer risk—namely, single nucleotide polymorphisms from a DNA repair gene, and smoking. Furthermore, we identified potentially heterogeneous groups of subjects characterized by distinct patterns of association. Cox proportional hazard models comparing clinical outcome variables between the cases of the two largest groups yielded a significant, meaningful difference in survival time in years (survivorship). A marginally significant difference in recurrence time was also noted. These results support the hypothesis that an LCS approach can offer greater insight into complex patterns of association.ConclusionsThis methodology appears to be well suited to the dissection of disease heterogeneity, a key component in the advancement of personalized medicine.

An Extended Michigan-Style Learning Classifier System for Flexible Supervised Learning, Classification, and Data Mining

“…AF speeds up effective learning by gradually guiding the algorithm to more intelligently explore reliable attribute patterns. These mechanisms and their application are further detailed in [7] and [8].…”

“…Attribute tracking (AT) is akin to longterm memory for supervised, iterative learning (see (8) in Figure 1). For a finite training dataset, a vector of accuracy scores is maintained for each instance in the data.…”

A multi-core parallelization strategy for statistical significance testing in learning classifier systems

“…Previously, we introduced a promising new methodology to address these complexities using a Learning Classifier System (LCS) algorithm [13]. Learning classifier systems (LCSs) [17] are a rule-based class of algorithms which combine machine learning with evolutionary computing and other heuristics to produce an adaptive system.…”

“…In [13] we applied our own extended supervised-learning classifier system, called AF-UCS [12], and a statistical and visualization guided knowledge discovery pipeline [14] to a real world genetic epidemiology study of bladder cancer susceptibility. As a result, we successfully replicated the identification of previously characterized factors that modify bladder cancer risk: i.e.…”

“…To date, formal significance testing of LCS run statistics, such as test accuracy, has been limited to the aforementioned study [13]. One of the major goals of the LCS research community is to extend these unique and powerful algorithms to real world applications.…”