Detecting, characterizing, and interpreting nonlinear gene-gene interactions using multifactor dimensionality reduction.

摘要: Human health is a complex process that dependent on many genes, environmental factors and chance events are perhaps not measurable with current technology or simply unknowable. Success in the design execution of population-based association studies to identify those genetic play an important role human disease will depend our ability embrace, rather ignore, complexity genotype phenotype mapping relationship for any given ecology. We review here three general computational challenges must be addressed. First, data mining machine learning methods needed model nonlinear interactions between multiple factors. Second, filter wrapper attribute large solution landscapes. Third, visualization help interpret models results. provide overview multifactor dimensionality reduction (MDR) method was developed addressing each these challenges.