Using the quantitative subphenotypes, SOC and BEH, we did not find significant linkage signals, even though we found reasonably high heritabilities for both traits (.35 for SOC and .52 for BEH). Ascertainment bias, where a pedigree was selected only if there were at least two individuals above thresholds for both the social and communication total scores, may have a large impact on heritability estimates (64). No ascertainment correction was performed since there is no efficient method for the complex ascertainment used in this study and no normative data from a general population sample exist for these two quantitative subphenotypes. In addition there may simply not be enough variation among the ASD patients to detect linkage. This is especially the case for SOC, where affected subjects had to reach a threshold to be included. In contrast, no such threshold exists for BEH. It is also important to remember that these domain total scores are summaries of the scores from many items that themselves may be heterogeneous and belong to different dimensions, so the quantitative traits we used may be as complex as the binary diagnosis of autism (65).
