CoChat
Sign Up
Cited by 249Genetics and Neurodevelopmental DisordersAutism Spectrum Disorder ResearchBioinformatics and Genomic Networks

The DOI matches its Crossref record; 53 of the 53 references it cites were found in a registry or an index.

The discovery of integrated gene networks for autism and related disorders

Open access
Hormozdiari, Fereydoun, Penn, Osnat, Borenstein, Elhanan, Eichler, Evan E.·2015·Genome Research
DOI 10.1101/gr.178855.114PMC PMC4317170
Open this paper in CoChat — free accountOpen PDF

Abstract

Despite considerable genetic heterogeneity underlying neurodevelopmental diseases, there is compelling evidence that many disease genes will map to a much smaller number of biological subnetworks. We developed a computational method, termed MAGI (m erging a ffected g enes into i ntegrated networks), that simultaneously integrates protein–protein interactions and RNA-seq expression profiles during brain development to discover “modules” enriched for de novo mutations in probands. We applied this method to recent exome sequencing of 1116 patients with autism and intellectual disability, discovering two distinct modules that differ in their properties and associated phenotypes. The first module consists of 80 genes associated with Wnt, Notch, SWI/SNF, and NCOR complexes and shows the highest expression early during embryonic development (8–16 post-conception weeks [pcw]). The second module consists of 24 genes associated with synaptic function, including long-term potentiation and calcium signaling with higher levels of postnatal expression. Patients with de novo mutations in these modules are more significantly intellectually impaired and carry more severe missense mutations when compared to probands with de novo mutations outside of these modules. We used our approach to define subsets of the network associated with higher functioning autism as well as greater severity with respect to IQ. Finally, we applied MAGI independently to epilepsy and schizophrenia exome sequencing cohorts and found significant overlap as well as expansion of these modules, suggesting a core set of integrated neurodevelopmental networks common to seemingly diverse human diseases.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1101/gr.178855.114

Cites

53 of 53 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1. Flowchart of MAGI. Given PPI and coexpression networks and case and control mutations, MAGI detects highly connected modules that are enriched for mutations in cases. The first phase calculates a score for each gene in the networks and selects seed pathways with high scores based on an extension of the color-coding algorithm (Alon et al. 1995). In the second phase, MAGI merges the seeds into modules using a random-walk approach and improves each one of them by applying a local search. The output consists of the best module detected, as well as a set of suboptimal modules. Each gene is assigned a “confidence score” according to its frequency within suboptimal modules.
Figure 1. Flowchart of MAGI. Given PPI and coexpression networks and case and control mutations, MAGI detects highly connected modules that are enriched for mutations in cases. The first phase calculates a score for each gene in the networks and selects seed pathways with high scores based on an extension of the color-coding algorithm (Alon et al. 1995). In the second phase, MAGI merges the seeds into modules using a random-walk approach and improves each one of them by applying a local search. The output consists of the best module detected, as well as a set of suboptimal modules. Each gene is assigned a “confidence score” according to its frequency within suboptimal modules.CC BY-NC · PubMed Central
Figure 2. Module M1. (A) Genes detected as part of module M1_Extended are displayed as graph nodes using Cytoscape (Shannon et al. 2003). Node colors reflect the score of each gene based on the number and type of de novo mutations: The more intense red color indicates a higher score, while gray indicates a score of zero (no de novo mutations observed). Edges (black lines) between two nodes represent genes that interact with each other according to the PPI network and are also highly coexpressed (Pearson correlation coefficient , i.e., the genes are included in the top 5% of gene pair coexpression during brain development). The innermost circle contains genes detected in > 99% of the suboptimal solutions. Subsequent concentric circles display genes found in > 80%, 20%, and 5% (M1_Extended) of the suboptimal solutions, respectively (see Methods). Nodes with black outlines are the ones detected in the optimal module detected (M1_Best). For a force-directed layout of this module, see Supplemental Figure 38. (B) The M1_Best score (dashed black line) shown in comparison to the top-scored module of 200 simulations using null model Null-1. (C) The number of LoF mutations covered by the top-scoring module (M1_Best) found using proband mutations versus the number of simulated LoF mutations covered by the top-scoring modules found under the same simulation. (D) The score of the top module found using siblings and control mutations (dashed black line) in comparison to the top-scored module of 200 simulations with the same number of mutations using Null-1. The siblings’ simulations were performed without using the ESP constraint (although similar results were obtained when an ESP constraint was applied). (E) The number of LoF mutations covered by the top-scoring module found using siblings’ mutations versus the number of simulated LoF mutations covered by the top-scoring modules found using Null-1. Sibling simulations were performed without filtering based on ESP controls.
Figure 2. Module M1. (A) Genes detected as part of module M1_Extended are displayed as graph nodes using Cytoscape (Shannon et al. 2003). Node colors reflect the score of each gene based on the number and type of de novo mutations: The more intense red color indicates a higher score, while gray indicates a score of zero (no de novo mutations observed). Edges (black lines) between two nodes represent genes that interact with each other according to the PPI network and are also highly coexpressed (Pearson correlation coefficient , i.e., the genes are included in the top 5% of gene pair coexpression during brain development). The innermost circle contains genes detected in > 99% of the suboptimal solutions. Subsequent concentric circles display genes found in > 80%, 20%, and 5% (M1_Extended) of the suboptimal solutions, respectively (see Methods). Nodes with black outlines are the ones detected in the optimal module detected (M1_Best). For a force-directed layout of this module, see Supplemental Figure 38. (B) The M1_Best score (dashed black line) shown in comparison to the top-scored module of 200 simulations using null model Null-1. (C) The number of LoF mutations covered by the top-scoring module (M1_Best) found using proband mutations versus the number of simulated LoF mutations covered by the top-scoring modules found under the same simulation. (D) The score of the top module found using siblings and control mutations (dashed black line) in comparison to the top-scored module of 200 simulations with the same number of mutations using Null-1. The siblings’ simulations were performed without using the ESP constraint (although similar results were obtained when an ESP constraint was applied). (E) The number of LoF mutations covered by the top-scoring module found using siblings’ mutations versus the number of simulated LoF mutations covered by the top-scoring modules found using Null-1. Sibling simulations were performed without filtering based on ESP controls.CC BY-NC · PubMed Central
Figure 3. Module M2. MAGI reiteration after M1 components removed. For details, see legend of Figure 2. For a force-directed layout of this module, see Supplemental Figure 39.
Figure 3. Module M2. MAGI reiteration after M1 components removed. For details, see legend of Figure 2. For a force-directed layout of this module, see Supplemental Figure 39.CC BY-NC · PubMed Central
Figure 4. Severity of missense mutations and differences in temporal patterns of expression. (A) The distribution of C-scores of probands’ missense mutations found inside M1_Extended or M2_Extended, outside M1_Extended and M2_Extended, and siblings' and controls' missense mutations. (B) Average normalized expression of all brain subtissues for M1_Extended and M2_Extended during brain development. Error bars represent mean ± SEM.
Figure 4. Severity of missense mutations and differences in temporal patterns of expression. (A) The distribution of C-scores of probands’ missense mutations found inside M1_Extended or M2_Extended, outside M1_Extended and M2_Extended, and siblings' and controls' missense mutations. (B) Average normalized expression of all brain subtissues for M1_Extended and M2_Extended during brain development. Error bars represent mean ± SEM.CC BY-NC · PubMed Central
Figure 5. Overlap of the modules identified for different neurodevelopmental diseases. (A) Venn diagram representing the overlap between genes that carry de novo mutations and are detected as part of the first modules when analyzing ASD without ID (IQ ≥ 70), ID with or without reported ASD (IQ < 70), and schizophrenia. Genes with LoF mutations are colored in red. Genes with only missense mutations are colored in black. Asterisks indicate genes for which mutations have been observed in two different groups. (B) Venn diagram representing the overlap between genes that carry de novo mutations and are detected as part of the second module of ID, the second module of schizophrenia, and the first module of epilepsy. The P-value reported for each disease is the maximum P-value of the three null models.
Figure 5. Overlap of the modules identified for different neurodevelopmental diseases. (A) Venn diagram representing the overlap between genes that carry de novo mutations and are detected as part of the first modules when analyzing ASD without ID (IQ ≥ 70), ID with or without reported ASD (IQ < 70), and schizophrenia. Genes with LoF mutations are colored in red. Genes with only missense mutations are colored in black. Asterisks indicate genes for which mutations have been observed in two different groups. (B) Venn diagram representing the overlap between genes that carry de novo mutations and are detected as part of the second module of ID, the second module of schizophrenia, and the first module of epilepsy. The P-value reported for each disease is the maximum P-value of the three null models.CC BY-NC · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 53 references

Cited by

Sorted by citations

See all 249

Related

Metadata

Authors
Hormozdiari, F., Penn, O., Borenstein, E., & Eichler, E. E.
Affiliations
  • University of Washington (US)
  • Howard Hughes Medical Institute (US)
ORCID
Venue
Genome Research
Publisher
Cold Spring Harbor Laboratory
Published
2014-11-05
Volume
25
Issue
1
Pages
142-154
ISSN
  • 1088-9051 (ISSN-L)
  • 1549-5469
DOI
10.1101/gr.178855.114
PMC
PMC4317170
OpenAlex
W2106870542
Cited by
249
References
53
Citations by year
2026: 5 · 2025: 11 · 2024: 11 · 2023: 16 · 2022: 20 · 2021: 20 · 2020: 29 · 2019: 27 · 2018: 23 · 2017: 27 · 2016: 43 · 2015: 16 · 2014: 1
Type
posted content
Language
en
Open access
Locations
  • Genome Research (publishedVersion)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
  • eScholarship (California Digital Library) (submittedVersion, cc-by-nc)
Funders
  • Simons Foundation (303241)
  • NIH/NIMH (R01MH101221)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetics and Neurodevelopmental Disorders
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Autism Spectrum Disorder Research
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
Keywords
autism spectrum disorder, intellectual disability, de novo mutations, gene regulatory networks, Three Kings, protein-protein interaction, RNA-seq expression, brain development, Wnt signaling, Notch signaling, SWI/SNF complex, synaptic function, long-term potentiation, calcium signaling, epilepsy, schizophrenia, neurodevelopmental disorders, missense mutations, disease severity
MeSH terms
Algorithms, Autistic Disorder, Epilepsy, Humans, Phenotype, Schizophrenia, Cohort Studies, Cluster Analysis, Databases, Factual, Sequence Analysis, RNA, Genetic Heterogeneity, Mutation, Missense, Gene Regulatory Networks, Exome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review