Autism Spectrum Disorder (ASD) is an early onset developmental disorder characterized by deficits in communication and social interaction and restrictive or repetitive behaviors 1,2 . Family studies demonstrate that ASD has a significant genetic basis with contributions both from inherited and de novo variants 3,4 . It has been estimated that de novo mutations may contribute to 30% of all Reprints and permissions information is available at www.nature.com/reprints.Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use:http:// www.nature.com/authors/editorial_policies/license.html#terms

Transcription factors (TFs) bind DNA by recognizing specific sequence motifs, typically of length 6–12bp. A motif can occur many thousands of times in the human genome, but only a subset of those sites are actually bound. Here we present a machine learning framework leveraging existing convolutional neural network architectures and model interpretation techniques to identify and interpret sequence context features most important for predicting whether a particular motif instance will be bound. We apply our framework to predict binding at motifs for 38 TFs in a lymphoblastoid cell line, score the importance of context sequences at base-pair resolution, and characterize context features most predictive of binding. We find that the choice of training data heavily influences classification accuracy and the relative importance of features such as open chromatin. Overall, our framework enables novel insights into features predictive of TF binding and is likely to inform future deep learning applications to interpret non-coding genetic variants.

Recent studies have demonstrated a strong contribution of germline de novo mutations to autism spectrum disorders (ASD). Tandem repeats (TRs), consisting of repeated sequence motifs of 1-20bp, comprise one of the largest sources of human genetic variation. Yet, the contribution of TR mutations to ASD has not been assessed on a genome-wide scale. Here, we leverage novel bioinformatics tools and~35 × whole genome sequencing of~1,600 families from the Simons Simplex Collection (SSC) to analyze germline de novo TR mutations at nearly 1 million loci in ASD-affected and unaffected siblings. We identify an average of 54 high-confidence mutations per child at an estimated true positive rate of 90%. We find novel genome-wide TR mutation patterns, including a bias toward larger mutations from the maternal compared with paternal germlines. We demonstrate a significant genome-wide excess of TR mutations in ASD probands, which are significantly larger than in controls, enriched in promoters of fetal brain expressed genes, and more strongly predicted to alter expression during brain development. Overall, our results indicate a significant, but so far overlooked, contribution of repetitive regions to ASD.