Research ArticleGENETICS

Systematic reconstruction of autism biology from massive genetic mutation profiles

See allHide authors and affiliations

Science Advances  11 Apr 2018:
Vol. 4, no. 4, e1701799
DOI: 10.1126/sciadv.1701799
  • Fig. 1 Multilevel comparison of DN mutations between or within ASD whole-exome studies.

    Venn diagrams and test statistics on overlap (A) between SSC and ASC ASD cohorts, and (B) between probands and siblings of SSC, at different levels. Correlation of pathway analysis statistics (C) between SSC and ASC ASD cohorts, and (D) between probands and siblings of SSC. The term “considered” or “selected” refers to items before or after the selection process at each level (Materials and Methods). See table S4 for full lists of selected variants and genes used in the analysis.

  • Fig. 2 Autism genetic association analysis across variant, gene, and pathway levels with the SSC exome mutation data.

    Columns 1 and 2: DN event rates and association test by gene- and pathway-level effects. Column 3: A descriptive model for autism genetic association. Rows are marginal (row 1) and conditional (rows 2 and 3) association tests and statistics, and corresponding model representations. Variants are grouped based on gene-level effects (G) [silent, missense, LGD, and nonsilent (missense + LGD)] and pathway-level effects (P) (hitting Selected pathways or Others). The associations marked in column 3 should be taken as the theme for corresponding row. Notations α(G; A) and α(G; A |P) are read as marginal association between G and A, and their conditional association given P. The association can be measured by rate difference (over noise), rate ratio (θ), or log θ. P values come from the rate difference tests. Error bars represent SEM. Note raw counts are summarized in table S1. For technical details, see the “Multilevel association analysis” section under Materials and Methods. NA, not applicable.

  • Fig. 3 An integrated view of autism-associated DN variants or genes from multiple sources in selected KEGG pathways.

    (A) hsa04310 Wnt signaling pathway, (B) hsa04727 GABAergic synapse, and (C) hsa04724 Glutamatergic synapse (next page). DN variant data come from SSC and ASC studies, and reported autism genes come from the SFARI Gene database. Gene-level scores (Materials and Methods) are marked by color. P values are from pathway analysis (Table 1). Data are integrated and visualized on KEGG pathway graphs using Pathview (23).

  • Fig. 4 Functional consequences of autism-associated missense mutations.

    (A) Ratios of damaging events as predicted by SIFT. (B) Ratios of events hitting a function domain defined in Pfam. (C) 1D protein domain structure and missense variants of all neurotransmitter receptors, transporters, and ion channel genes in synapse pathways (the bold black box nodes in pathway graphs in figs. S6 and S7). (D) 1D and 3D protein structures and missense variants hitting the AC, that is, ADCY5, and interacting G proteins, GNAS (Gs) and GNAO1 (Gi/o). The pathway context is shown in the green dashed box in fig. S7. AC controls the production of cAMP second-messenger in synapses (section S4). Error bars represent SEM.

  • Fig. 5 The superpathway clusters emerged from the pathway-level analysis of SSC exome variant data.

    Seven selected pathways are highly interconnected and frequently connected to five additional pathways. All these pathways form a superpathway-level network. Two clusters or modules emerged with distinct topology and function.

  • Table 1 Significant pathways selected from SSC and ASC exome data.

    Columns P and q are the P- and q-values from marginal tests, and Pcon and qcon are from the conditional tests. Columns j and k are the marginal and conditional counts of selected genes. These pathways are likely drivers or disease-causing pathways because of the special analysis procedure (Materials and Methods). See table S4 for full lists of selected variants and genes.

    PqPconqconjkReference
    SSC pathways
      hsa00310 Lysine degradation8.0 × 10−50.028.0 × 10−50.0288(11, 44)
      hsa04727 GABAergic synapse4.9 × 10−30.313.9 × 10−30.0388(45, 46)
      hsa04974 Protein digestion and absorption1.5 × 10−20.359.7 × 10−30.0777(35)
      hsa04310 Wnt signaling pathway7.9 × 10−30.318.5 × 10−30.08109(47, 48)
      hsa04810 Regulation of actin cytoskeleton2.5 × 10−20.356.3 × 10−30.061212(49, 50)
      hsa04010 MAPK signaling pathway2.1 × 10−20.359.1 × 10−30.071410(51, 52)
      hsa04530 Tight junction1.7 × 10−20.356.6 × 10−20.1795(53, 54)
      hsa04713 Circadian entrainment7.7 × 10−30.316.6 × 10−20.1483(34, 55)
      hsa04724 Glutamatergic synapse2.1 × 10−20.351.9 × 10−10.1982(56, 57)
    ASC pathways
      hsa00310 Lysine degradation6.2 × 10−50.016.2 × 10−50.0155(11, 44)
      hsa04713 Circadian entrainment1.5 × 10−30.159.9 × 10−40.0155(34, 55)
      hsa04976 Bile secretion3.5 × 10−30.188.4 × 10−30.0643(35)
      hsa04727 GABAergic synapse7.8 × 10−30.214.4 × 10−20.1242(45, 46)
      hsa04010 MAPK signaling pathway7.1 × 10−30.216.7 × 10−20.0774(51, 52)

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/4/e1701799/DC1

    section S1. Sequential convergence from variant to gene to pathway level

    section S2. Autism genetic association dissected across multiple levels

    section S3. Pathways of DN events, integrated molecular mechanism

    section S4. Subpathway biology, coherent fine details

    section S5. Superpathway biology, emergent big picture

    fig. S1. Multilevel recurrence of de novo mutations in SSC study.

    fig. S2. Autism genetic association analysis across variant, gene, and pathway levels with the SSC exome DN mutation data.

    fig. S3. Prevalence of DN variant types by gene- and pathway-level effects in the SSC exome study.

    fig. S4. An integrated view of autism associated DN variants or genes from multiple sources.

    fig. S5. An integrated view of DN variants by gene level effects in the selected KEGG pathway, hsa04310 Wnt signaling pathway.

    fig. S6. An integrated view of DN variants by gene level effects in the selected KEGG pathways, hsa04727 GABAergic synapse.

    fig. S7. An integrated view of DN variants by gene level effects in the selected KEGG pathways, hsa04724 Glutamatergic synapse.

    fig. S8. 1D protein domain structure and missense variants of genes in Wnt signaling pathway for both probands and siblings.

    fig. S9. 1D protein domain structure and missense variants of genes in synapse pathways for both probands and siblings.

    fig. S10. 1D protein domain structure and DN mutations of gene pairs in Wnt signaling and synapse pathways.

    fig. S11. 1D and 3D protein structures and missense variants hitting mGluR inhibitor GRK (ADRBK2) and interacting G proteins.

    fig. S12. Selected GO terms and emerging biological themes from the SSC exome variant data.

    table S1. The actual (and expected) counts of DN events by gene-level (columns) and pathway-level (rows) effects.

    table S2. Other pathways connected by three or more selected pathways in Table 1.

    table S3. Significant GO groups selected from SSC, their test statistics, and references.

    table S4. (selectVariant.xlsx file). Validated and selected variants in the multilevel integrated analyses of SSC and ASC data.

    table S5. (variantAnnot.xlsx file). Validated variants and functional annotations in the selected pathways of SSC data.

    table S6. (geneAnnot.xlsx file). Genes hit by LGD or missense events in probands in selected pathways.

    References (5889)

  • Supplementary Materials

    This PDF file includes:

    • section S1. Sequential convergence from variant to gene to pathway level
    • section S2. Autism genetic association dissected across multiple levels
    • section S3. Pathways of DN events, integrated molecular mechanism
    • section S4. Subpathway biology, coherent fine details
    • section S5. Superpathway biology, emergent big picture
    • fig. S1. Multilevel recurrence of de novo mutations in SSC study.
    • fig. S2. Autism genetic association analysis across variant, gene, and pathway levels with the SSC exome DN mutation data.
    • fig. S3. Prevalence of DN variant types by gene- and pathway-level effects in the SSC exome study.
    • fig. S4. An integrated view of autism associated DN variants or genes from multiple sources.
    • fig. S5. An integrated view of DN variants by gene level effects in the selected KEGG pathway, hsa04310 Wnt signaling pathway.
    • fig. S6. An integrated view of DN variants by gene level effects in the selected KEGG pathways, hsa04727 GABAergic synapse.
    • fig. S7. An integrated view of DN variants by gene level effects in the selected KEGG pathways, hsa04724 Glutamatergic synapse.
    • fig. S8. 1D protein domain structure and missense variants of genes in Wnt signaling pathway for both probands and siblings.
    • fig. S9. 1D protein domain structure and missense variants of genes in synapse pathways for both probands and siblings.
    • fig. S10. 1D protein domain structure and DN mutations of gene pairs in Wnt signaling and synapse pathways.
    • fig. S11. 1D and 3D protein structures and missense variants hitting mGluR inhibitor GRK (ADRBK2) and interacting G proteins.
    • fig. S12. Selected GO terms and emerging biological themes from the SSC exome variant data.
    • table S1. The actual (and expected) counts of DN events by gene-level (columns) and pathway-level (rows) effects.
    • table S2. Other pathways connected by three or more selected pathways in Table 1.
    • table S3. Significant GO groups selected from SSC, their test statistics, and references.
    • Legends for tables S4 to S6
    • References (58–89)

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • table S4 (Microsoft Excel format). Validated and selected variants in the multilevel integrated analyses of SSC and ASC data.
    • table S5 (Microsoft Excel format). Validated variants and functional annotations in the selected pathways of SSC data.
    • table S6 (Microsoft Excel format). Genes hit by LGD or missense events in probands in selected pathways.

    Files in this Data Supplement:

Navigate This Article