Predicting Autism Phenotypic Outcomes Based On Genotypic Information

Voiceover

Presenter(s)

Eleanor Sams

Abstract or Description

Characterizing the structure and function of the human genome in association with phenotypic presentation at the molecular and organism level is critical for understanding human health and disease. Prediction of phenotypes from genotypes is an area of ongoing investigation in the field of genomics. In this study, we constructed a matrix containing known genetic and phenotypic information for a cohort of individuals to use in a predictive model for genotype-phenotype associations. The genetic event used in this model is the presence of a copy number variant (CNV) in the 16p11.2 region, and the associated phenotype is autism. The relevant data was obtained from the Simons Searchlight 16p11.2 cohort and contains individuals with autism and their affected/unaffected family members. To build the genotype section of the matrix, the copy number for each of the 52 genes in the 16p11.2 region was filled in for each sample using CNV coordinates for events confirmed by two technologies (microarray and whole exome sequencing). Every sample in the matrix also received a score based on if they have autism and any related phenotypes. To expand the ideas used in designing this 16p11.2 model, a genome-wide matrix was built such that the copy number for every gene in the genome was recorded for each sample. Both matrices were statistically analyzed to determine the decisive factors in distinguishing between individuals with and without autism. This analysis revealed that the related phenotypes are most important in the autism diagnosis distinction for the 16p11.2 matrix, whereas genotypic differences are most critical for the genome-wide matrix. Analysis of both matrices revealed that autism diagnosis is more commonly associated with 16p11.2 deletions. Based on our results, it is our hope that this prototype model can be utilized for additional phenotypes in the clinic.