ABSTRACT Substantial investments are being made to sequence the genomes of families with autism and other neurodevelopmental disorders (NDD). However, identifying disease mutations outside the ~1% of protein coding sequences is challenging because 1) the ‘search space’ is much larger, and thus many more mutations occur by chance, and 2) there is no simple code to identify deleterious mutations in non-coding sequences, and thus loss of function mutations must be defined experimentally. In addition, the consequences of mutations in non-coding (i.e. regulatory) sequences are often highly dependent on the specific cell type. Thus, functional assays must be conducted in vivo, in the appropriate CNS cell types. To address the search space challenge, we have focused specifically on the untranslated regions (UTRs) of mRNAs. UTRs are important, conserved regulatory sequences that profoundly impact protein levels by altering translation rates or transcript stability for specific genes. To address the lack of a code for interpreting UTR mutations, we have developed a unique combination of expertise to conduct massively parallel functional analysis of UTR variants from NDD patients, in relevant cell types in the brain. Combining two innovative but established components: massively parallel reporter assays, and cell type specific translational profiling, we aim to establish a pipeline to 1) Identify UTR mutations that result in altered protein levels, 2) conduct genetic burden and association testing on these variants, and 3) train machine learning models that can predict the effects of future mutations in developing neurons of the brain. This project will leverage the existing large investment in NDD genome sequencing by defining individual non-coding, disease- causing mutations in a class of sequences that has, so far, not been the focus of disease studies.