Delineating autism subtypes by phenotype-wide scan across genome-wide genotypes in a patient centric information commons

This grant was for a one-year study that would result in a phenome-wide scan across all genotypes measured by SNP array. The hope was that this would generate novel insights about the substructure of the different autisms. The involvement of genetic factors in ASD is demonstrated. Several genome-wide association studies (GWAS) of common single nucleotide polymorphisms have been performed but the effect sizes remained modest. New approaches, with a better integration of phenotypes and genotypes, such as pathway and network analyses could help to unravel the genetic mechanisms of ASD.  Stessman et al. suggested a “genotype-first” approach. In this approach, the selection criterion is no more phenotypic but genotypic: the variants of an identified gene of interest. Then, systematic associations of phenotypes with the variants of this gene are assessed. This approach could allow the discovery of new subtypes of ASD. This genotype-first approach is similar to another method described by Denny et al: Phenome-wide association studies (PheWAS). Dr. Kohane’s group demonstrated in previous work on thiopurine methyl-transferase enzymatic activity in the field of thiopurine therapy that this method could help to describe new subgroups of patients with specific characteristics. The PheWAS approach might allow the linkage of genes variants to specific sub-phenotypes of ASD. Some of these subgroups could benefit of a specific therapy, given that an earlier treatment can improve the situation. This kind of study requires large amounts of genotypic and phenotypic data. Big cohorts of ASD patients and families exist. They gather genetic and phenotypic data from thousands of patients, representing a promising source of data. One of the issues preventing wide research programs over these cohorts is the heterogeneity of the assessment tools for ASD phenotyping. Dr. Kohane argues that integrating genotypic and phenotypic data these cohorts into a single patient centric information platform enabling the unification of the different sources of data would allow a more effective use of their data for research purposes. One of the challenges preventing an effective use of this knowledge gold mine is the fragmentation of the data. Indeed, it is difficult to analyze phenotypic data fragmented over several different tools, representing thousands of questions and answers that may or may not overlap between the tools. Dr. Kohane’s group thinks that the unification and the harmonization of all these phenotypic data into single concept based ontology might enable its effective use in research. Collecting phenotypic data is expensive and time consuming. Lots of phenotypic data are collected on a daily basis but are difficult to access for research purposes. Therefore, clinical data warehouses (CDW), like i2b2, were designed to enable second use of data collected for health care for research purposes. The Children’s Hospital Boston (CHB) is equipped with such CDW and a significant part of ASD children from SSC are also treated in this hospital. With this grant, Dr. Kohane’s group integrated anonymized data from CHB CDW to SSC phenotypic data, augmenting the number of phenotypic data available for analyses by performing a phenotypic expansion.