With individual SNPs explaining minimal variance in risk for complex traits, a complementary approach is to combine SNPs of smaller effect in order to create a polygenic risk score to test for association with a phenotype of interest. A common polygenic score approach involves weighting alleles based on their association in a training sample and then summing across a subset of SNPs to create a score for each individual within a second sample (Purcell et al., 2009; Wray et al., 2014). One such study, published for externalizing behaviors, found that while polygenic scores cross-predicted between an adult and adolescent/young-adult sample, they still only accounted for a small amount of the variance (Salvatore et al., 2015). Thus, despite some progress in gene finding efforts using a variety of methodologies, the adolescent behavior field would benefit from large-scale, coordinated efforts to identify genes involved in key adolescent outcomes. Some of these are underway for phenotypes such as adolescent aggression (The Eagle Consortium, Pappa et al., 2015).
