Identifying Windows of Susceptibility by Temporal Gene Analysis

Kristin P. Bennett, Elisabeth M. Brown, Hannah De los Santos, Matthew Poegel, Thomas R. Kiehl, Evan W. Patton, Spencer Norris, Sally Temple, John Erickson, Deborah L. McGuinness, Nathan C. BolesScientific 2019 · DOI

TL;DR

This paper introduces the "Susceptibility Windows Ontological Transcriptome" (SWOT) clock, a novel computational tool that integrates temporal gene expression data with disease and pathway databases to pinpoint critical developmental periods ("windows of susceptibility") for neurodevelopmental disorders. This approach offers a more precise and context-aware method for identifying when environmental or biological insults might lead to disease, paving the way for better preventative strategies and interventions.

What Problem Does This Paper Address?

Understanding neurodevelopmental disorders is complicated by the fact that genetic mutations, environmental toxins, and other biological insults often exert their effects during specific, vulnerable periods in early development. Identifying these "windows of susceptibility" (WOS) is crucial for both protecting developing fetuses and designing timely interventions. However, current methods for identifying WOS are often experimental, time-consuming, and costly. While computational approaches exist for analyzing gene expression and identifying disease associations, they often lack a developmental context, making it difficult to pinpoint when these disruptions might occur. This paper aims to bridge this gap by developing a computational framework that integrates temporal gene expression data with existing biological knowledge to predict WOS for various neurodevelopmental disorders.

Methods

The core of this work is the development of the Susceptibility Windows Ontological Transcriptome (SWOT) Clock. The process begins with:

  • Data Acquisition and Preprocessing: Utilizing temporal gene expression data from human pluripotent stem cells (hPSCs) undergoing cortical development (CORTECON dataset). This data is normalized and scaled.
  • Singular Value Decomposition (SVD): SVD is applied to the gene expression data to reduce dimensionality and identify key temporal patterns (eigengenes and eigendays) that capture the overall developmental trajectory. This helps to represent the complex temporal dynamics in a more manageable way.
  • Clustering Analysis: Fuzzy C-Means clustering is used to group genes with similar temporal expression profiles. This allows for the identification of distinct molecular signatures associated with different developmental stages.
  • Semantic Integration and Enrichment Analysis: The identified gene clusters are then linked to external databases (e.g., Gene Ontology, disease-gene association databases like DISEASES, pathway databases like KEGG, and protein-protein interaction databases like STRINGDB) using a semantic framework (SemNExT). Enrichment analysis is performed for each developmental stage within each cluster to identify which stages are significantly associated with specific diseases or pathways.
  • SWOT Clock Visualization: The results are visualized as interactive "SWOT Clocks." These clocks display gene expression over developmental time, with connections between genes representing protein-protein interactions. The clocks highlight developmental stages with significant enrichment (positive Log Odds Ratio and low p-value) for disease-associated genes, indicating potential WOS.

Datasets

The primary dataset used is the CORTECON dataset, which provides temporal RNA-seq expression data from an in vitro model of human cortical development using hPSCs, covering nine distinct time points.

Additionally, the study leverages several publicly available databases for enrichment and association analysis:

  • DISEASES database: For disease-gene associations.
  • SFARI Gene database: For autism candidate genes.
  • Literature-derived gene lists: For autism and schizophrenia from published studies.
  • KEGG Pathways: For pathway enrichment analysis.
  • STRINGDB: For protein-protein interaction data.
  • Online Mendelian Inheritance in Man (OMIM): For disease-gene associations.

Key Insights and Ideas

  • Developmental Context is Key: The most significant insight is that understanding disease susceptibility requires a temporal, developmental perspective. Simply identifying genes associated with a disease is insufficient; knowing when those genes are active and vulnerable during development is critical.
  • SWOT Clock as an Integrative Tool: The SWOT clock effectively integrates diverse data types (temporal gene expression, protein interactions, disease associations, pathways) into a single, interpretable visualization, bridging the gap between raw data and biological understanding.
  • Predicting WOS for Multiple Disorders: The method successfully identified potential WOS for a range of neurodevelopmental disorders, including autism spectrum disorder, schizophrenia, and microcephaly, demonstrating its broad applicability.
  • Shared and Distinct Mechanisms: The analysis revealed both shared and distinct WOS for related disorders (e.g., ASD and schizophrenia), suggesting common underlying mechanisms but also highlighting unique developmental vulnerabilities.
  • Hypothesis Generation Power: The SWOT clock serves as a powerful hypothesis-generation tool, identifying potential WOS and pathways that warrant further experimental investigation.

Key Results

  • The study identified specific developmental stages as potential windows of susceptibility for various neurodevelopmental disorders. For example:
  • Autism Spectrum Disorder (ASD): Identified Neural Differentiation (ND), Cortical Specification (CS), and Upper Layers (UL) stages as potential WOS.
  • Schizophrenia: Identified the Upper Layers (UL) stage as a potential WOS.
  • Microcephaly: Identified the Neuroectoderm (NE) stage as a potential WOS.
  • The analysis of Zika virus (ZIKV) infection revealed an early WOS during the Neuroectoderm (NE) stage for ZIKV-induced microcephaly, with distinct patterns for down-regulated and up-regulated genes.
  • The study demonstrated that the SWOT clock could identify WOS even when using different types of gene lists (e.g., from literature, candidate genes, or differential expression) and experimental designs (e.g., adult brain studies vs. organoid models).

Limitations and Future Work

  • Data Quality Dependence: The accuracy of the WOS predictions is heavily dependent on the quality and comprehensiveness of the input temporal gene expression data. Improvements in identifying differentially expressed genes in time-course data could enhance accuracy.
  • Correlative Nature: The current approach identifies associations and WOS but does not establish underlying causal mechanisms. Further experimental validation is required to confirm these hypotheses.
  • Single in vitro Model: The primary CORTECON dataset is derived from a single in vitro model of human development. Expanding the analysis to include data from a broader range of sources and in vivo models would strengthen the findings.
  • Future Work: The authors suggest several avenues for future research:
  • Developing meta-analysis approaches and adjusting for individual dataset uncertainties.
  • Collating more diverse datasets and creating a more comprehensive tool.
  • Applying the SWOT clock to other developmental disorders and organ systems.
  • Further investigating the identified pathways and gene networks to elucidate underlying mechanisms.
  • The SWOT clock tool is available for exploration and collaboration.

Summary updated

Abstract

Increased understanding of developmental disorders of the brain has shown that genetic mutations, environmental toxins and biological insults typically act during developmental windows of susceptibility. Identifying these vulnerable periods is a necessary and vital step for safeguarding women and their fetuses against disease causing agents during pregnancy and for developing timely interventions and treatments for neurodevelopmental disorders. We analyzed developmental time-course gene expression data derived from human pluripotent stem cells, with disease association, pathway, and protein interaction databases to identify windows of disease susceptibility during development and the time periods for productive interventions. The results are displayed as interactive Susceptibility Windows Ontological Transcriptome (SWOT) Clocks illustrating disease susceptibility over developmental time. Using this method, we determine the likely windows of susceptibility for multiple neurological disorders using known disease associated genes and genes derived from RNA-sequencing studies including autism spectrum disorder, schizophrenia, and Zika virus induced microcephaly. SWOT clocks provide a valuable tool for integrating data from multiple databases in a developmental context with data generated from next-generation sequencing to help identify windows of susceptibility.
BibTeX
@inproceedings{bennett2019identifying,
  author = {Kristin P. Bennett and Elisabeth M. Brown and Hannah De los Santos and Matthew Poegel and Thomas R. Kiehl and Evan W. Patton and Spencer Norris and Sally Temple and John Erickson and Deborah L. McGuinness and Nathan C. Boles},
  title = {{Identifying Windows of Susceptibility by Temporal Gene Analysis}},
  booktitle = {Scientific},
  year = {2019},
  doi = {10.1038/s41598-019-39318-8},
}
Figure