Other analytical options include the statistical model (e.g. for a binary outcome, whether it is treated as simply yes/no or the time-to-event is also taken into account), the use of any covariates (e.g. for age, gender, or topic-specific variables), and the use of corrections for relatedness. Usually, the impact of these options is not major, but it can make a difference for borderline associations which may seem to pass or not pass a desired p-value threshold. This means that both for GWA studies and subsequent investigations, one should carefully report the methods in sufficient detail so they can be independently replicated by other researchers. [41]
