25980
Genome-Wide Association Study Suggests Genetic Homogeneity within Complex Autism Subgroup
Autism is phenotypically and genetically heterogeneous, prompting our focus on identifying, defining, and studying genotypic differences between clinical autism subgroups. Few studies have examined genetic differences between subgroups on a genome-wide scale. Furthermore, though the development of autism can seldom be explained by a single factor, little is known about how genetic factors interrelate to cause autism.
Objectives:
Since a large proportion of autism heritability is thought to be caused by common genetic variants (Gaugler et al., 2014), we expect that many cases of autism arise from specific combinations of common variants. We aim to discover associations between common variants and clinical autism subgroups. In particular, we focus on testing combinations of multiple variants to identify potential interactions that contribute to the development of autism. Examining all combinations of millions of variants is impossible, so we generate candidate combinations using a data-driven method that measures the prevalence of minor alleles in autism subgroups.
Methods:
Using the SFARI SSC dataset containing ~3 million variants, ASD probands were sorted as “essential” (n=436) or “complex” (n=76). Using a well-studied autism subtype classification, individuals are designated “complex” based on physical evidence of an insult to early morphogenesis (Miles, et al., 2008, Tammimies et al., 2015). Frequent pattern mining, a data mining algorithm, was used to calculate prevalence of variant combinations within each subgroup. We identified the variant combinations that had the highest difference in subgroup prevalence; these were tested for association with the subgroups.
Results:
After excluding combinations exhibiting linkage disequilibrium due to physical proximity, frequent pattern mining identified 14 individual variants and 27 combinations of variants that were at least twice as prevalent in the complex subgroup, versus 13 individual variants in the essential subgroup. 8 of the individual variants and 23 variant pairs were significantly associated with the complex subgroup (family-based association test; p<.01). In contrast, the essential subgroup had no associated variants, individually or in combination. We speculate that the complex subgroup is a more genotypically homogeneous group, leading to these stronger associations. We found multiple variants within the LPPR3 gene to be associated with the complex subgroup; to our knowledge this gene was not previously associated with autism. 16 of the 23 significant variant pairs involved the ISM1 gene, linking it to genes and non-genic regions on various chromosomes. Stewart, et al. (2013) previously associated ISM1 with obsessive-compulsive disorder and noted several ISM1 gene-gene correlations related to that disorder as well, but the gene has not been specifically associated with autism.
Conclusions:
Our preliminary study identified several combinations of common variants associated with the complex dysmorphology autism subgroup. This suggests that the complex subtype is more genetically homogeneous than the essential subtype, contrary to prior belief. Further analysis is required to study the relationships between the implicated genes and how they might contribute to autism development. We expect that applying this method to more autism subtypes will lead to the discovery of more genetic distinctions between groups.