Genome-wide Association Mapping in Multiplex Autism Families. L. A. Weiss1, D. E. Arking2, T. Green1, J. F. Gusella1, S. L. Santangelo1, R. E. Tanzi1, P. Sklar1, A. Chakravarti2, M. J. Daly1, Autism Consortium Gene-Finding Group & Autism Genome Project 1) Ctr Human Genetic Research, Harvard-MGH/Broad, Boston, MA; 2) Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD.

   Autism is a highly heritable developmental disorder, though to date, identifying specific genes has met with limited success. We therefore initiated the first genome-wide association mapping study using 500,000 SNP markers in multiplex autism families. (Linkage analysis is reported in an abstract by Arking et al.) Our primary datasets were ~800 AGRE families genotyped on Affymetrix 5.0 at the Broad Institute, and an NIMH autism sample (250 families independent from AGRE), genotyped on the Affymetrix 500K/5.0 at Johns Hopkins University. Combining these datasets using stringent QC measures yielded 864 complete families with 1,594 trios. Data analysis was performed using PLINK. Initial TDT analysis did not yield genome-wide significant association. However, we utilized additional SNP microarray follow-up data, as well as designing Sequenom assays in our top hits for genotyping by the Autism Genome Project. In the combined dataset, an array SNP on chromosome 5p15 shows genome-wide significant association with autism. We followed up these results by performing imputation analysis to capture additional variation, identifying signals of interest that were validated by Sequenom genotyping. We also looked at positional and functional candidate regions previously implicated for autism, such as genes for Mendelian disorders associated with autism, genes containing rare variants associated with autism, regions of copy number variants associated with autism, and candidate genes for association between common variation and autism. The only region meeting criteria for region-wide association is the Williams Syndrome region on chromosome 7. Our results suggest that individual common variants of major effect do not explain a large proportion of the heritability of autism, nor do previously studied candidate genes. We report here a novel association of genetic variation on 5p15 with autism. This signal falls between SEMA5A and TAS2R1.