One advantage of this approach is that alleles that were present at low frequency in the DGRP and could not be detected by GWA can be represented at intermediate frequencies in the base population used to generate the advanced intercross. In addition, R428 cost extensive recombination generates a vast number of outbred
individuals so that sample size in the advanced intercross population is no longer limiting. Finally, changes in allele frequencies that occur during many (>25) generations of intercrossing can result in changes in additive effects of single variants that participate in gene–gene interactions, enabling significant associations to be uncovered in the extreme QTL mapping population that were not identified in the original GWA study in the DGRP [ 17•• and 18]. Combining the results from GWA analyses and extreme QTL mapping studies can reveal comprehensive genetic networks that underlie variation in the behavioral phenotype ( Figure 3). A number of generally applicable insights have emerged from these studies: First, most behavioral phenotypes are sexually dimorphic, implying distinct genetic architectures for males and females. Second, epistasis dominates the genetic architecture of complex traits, including behaviors [17••, 18, 39 and 40], and BIRB 796 suppressing epistasis
buffers the genome against the effects of newly arising mutations [39 and 40]. Third, common alleles have small to moderate effects on phenotypic variation, whereas rare alleles, that have perhaps appeared in more recent evolution, tend to have large effects [41 and 42]. Fourth, the genes that contribute to variation in behaviors are pleiotropic and span a wide range of gene ontology categories; however, developmental genes and genes associated
with neural connectivity and neuronal function are prominently represented among diverse behavioral phenotypes [17•• and 28]. This is perhaps not surprising as the expression of behaviors is itself a property of the nervous system. Since behaviors encompass interactions between organisms and their environments, the relationship between the genome and organismal phenotype is not static, but the genetic networks that orchestrate Montelukast Sodium the behavioral phenotype are expected to be dynamic and plastic. Examination of whole genome transcriptional profiles of an DGRP-derived advanced intercross population using Affymetrix expression microarrays under 20 different environments showed that only ∼15% of the transcriptome is environmentally plastic to macro-environmental changes, encompassing among others proteases and rapidly evolving multigene families [13••]. The remainder of the transcriptome is remarkably buffered (canalized) against environmental perturbations. Different genotypes can respond differently to environmental changes, which is the definition of ‘genotype-by-environment interactions’.