Through the Simons Simplex Collection (SSC) and the Autism Sequencing Consortium (ASC) sequencing efforts, thousands of de novo coding mutations have been identified in individuals with autism spectrum disorder (ASD), the majority of which are de novo missense (dnMis) mutations. Thus far, however, these mutations have not contributed much to our understanding of ASD, largely because their functional impact is unknown, unlike de novo loss-of-function mutations.

With funding support from a previous SFARI award, Haiyuan Yu, Bernie Devlin and Kathryn Roeder successfully implemented a novel interactome perturbation screening pipeline that integrates Clone-seq, an innovative massively parallel site-directed mutagenesis pipeline, with high-throughput green fluorescent protein (GFP) and yeast two-hybrid (Y2H) assays to study dnMis mutations associated with ASD¹.

Yu, Devlin and Roeder now plan to vastly scale up this pipeline in order to clone and examine >1,000 dnMis mutations identified from the SSC and SPARK cohorts. They will further improve the interactome perturbation screening pipeline by adding a mass spectrometry assay. This assay, which involves stable isotope labeling with amino acids in cell culture (SILAC) followed by immunoprecipitation and mass spectroscopy, can be applied to examine any mutation in any protein and detect both loss and gain of interactions at the whole proteome level, overcoming many limitations of Y2H assays.

Using their experimental data, they will also develop a novel multilayer random-forest-based machine learning pipeline, DENIM (DE Novo missense mutation IMpact), to predict the impact of all dnMis mutations on protein stability and interactions. They will further develop the widely-used statistical framework, TADA (Transmission And De novo Association)², to integrate the experimental results and DENIM predictions in order to identify novel ASD risk genes. They will then use DAWN (Detecting Association With Networks)³ and related networking tools to evaluate dynamic network data to determine subnetworks (i.e., communities) where unusual clustering of risk genes occur.

This experimentally and computationally integrated pipeline is expected to substantially increase the number of autism risk genes identified. It will also further our understanding of the interrelationships among risk genes and will determine the potential functional impact of mutations in these genes on ASD pathophysiology.