About ASD

ASD@Princeton is a web-interface for exploring autism-associated genes, created by the Laboratory for Bioinformatics and Functional Genomics in the Lewis-Sigler Institute for Integrative Genomics at Princeton University.

Through this interactive portal, we present our (1) genome-wide autism-gene predictions along with (2) cellular processes/pathways and (3) spatiotemporal brain signatures linked to ASD. Also presented is (4) prioritization of genes within CNV intervals linked to autism.

The interface allows gene-based search, dynamic visualization in the context of the human brain-specific gene network, and ability to export results.


  1. Why do we need to predict novel genes associated with autism?

    Several lines of evidence point to the fact that autism has a strong genetic basis, with several hundreds to over a thousand estimated potentially causal genes. However, currently (mid 2016) we only know about 65 such genes, most of which are targets of rare mutations. Therefore, in order to gain a more complete picture of the genetic basis of autism and map out the dysregulated cellular/developmental functions of the brain underlying the disease, we need to discover the full complement of autism-associated genes.

  2. What about previous work in this area?

    Several previous studies have used molecular interaction networks to characterize known experimentally determined autism spectrum disorder (ASD) genes (Willsey et al., 2013, Parikshak et al., 2013, Uddin et al., 2014, Xu et al., 2014, Li et al., 2014, Liu et al., 2014, Chang et al., 2015, Hormozdiari et al., 2015), very little effort has gone into predicting entirely novel ASD genes, and none of the previous methods have been able to make ASD-gene predictions across the genome. Furthermore, such an effort requires a genome-wide functional network focused on the brain, whereas networks such as protein-protein interactions (PPI) are largely incomplete, and PPI as well as integrated functional networks (until recently) lack tissue or cell-type specificity that is crucial for understanding disease genetics.

  3. How are the autism-gene predictions made here?

    We use a machine learning classifier that learns the connectivity patterns of known ASD genes in this brainspecific network and then uses these data-driven signals specific to ASD-associated genes to predict the level of potential ASD association for every gene in the genome. Using this approach, here we provide predictions of ASD-associated genes, including candidate genes that have minimal or no prior genetic evidence. Many of these candidate genes have been validated by sequencing studies as bona fide ASD-associated genes since the initial predictions were made.

  4. What is this 'brain-specific' network?

    The human brain-specific gene network that underlies our predictions was built using a Bayesian method that extracts and integrates brain-specific functional signals from thousands of gene expression, protein-protein interaction, and regulatory-sequence datasets. This integration creates a genome-wide probabilistic graph representing how proteins function together in pathways in the brain, or, intuitively, a molecular-level functional map of the brain. See Greene, Krishnan, Wong, et al. for more details on this network.

  5. How can I cite this resource?

    Please cite: Krishnan A*, Zhang R*, Yao V, Theesfeld CL, Wong AK, Tadych A, Volfovsky N, Packer A, Lash A, Troyanskaya OG. (2016) Genome-wide prediction and functional characterization of the genetic basis of autism spectrum disorder. Nature Neuroscience