Table 1 Summary of strategies to detect gene-gene and gene-environment interactions
Analytical strategy | Advantages | Disadvantages | References |
|---|---|---|---|
Examine the effect of the cumulative number of risk alleles at multiple loci | Simple; shows independence of loci | No interaction measured | |
Compare effect of risk allele in sample subgrouped by environmental exposure or additional genotype | Simple | Substantial loss of power in subgroups | |
Identify risk allele whose association with phenotype is modulated by inclusion of environmental or genetic covariate | Easy to implement | Multiple comparisons | |
Inclusion of interaction term in regression model | Direct modeling of gene-gene or gene-environment interaction | Need to define multiple terms in model; possibility of over-fitting; multiple comparisons | |
Non-linear statistical classification techniques, including Bayesian networks, neural networks and support vector machines | Large volumes of data in model-free manner | Difficult to interpret; require validation datasets |