NETMAGE: a humaN-disEase phenoType MAp GEnerator for the Visualization of PheWAS

Vivek Sriram, Manu Shivakumar, Sang‐Hyuk Jung, Lisa Bang, Anurag Verma, Seunggeun Lee, Eun Kyung Choe, Dokyoon KimbioRxiv 2020 · DOI

TL;DR

NETMAGE is a new web-based tool that visualizes complex genetic associations between diseases as interactive networks. This allows researchers to more easily explore potential pleiotropy (when a single gene affects multiple traits) and identify new drug targets for personalized medicine.

What Problem Does This Paper Address?

Understanding the genetic underpinnings of diseases is a monumental task. Phenome-Wide Association Studies (PheWAS) are powerful tools that link genetic variants to a wide range of phenotypic traits (diseases, medical conditions, etc.). However, interpreting the sheer volume of associations generated by PheWAS can be overwhelming. Specifically, identifying instances where a single gene influences multiple, seemingly unrelated, phenotypic traits (a phenomenon known as pleiotropy) is challenging with traditional data analysis methods. This lack of accessible visualization hinders researchers from gaining a deeper understanding of the genetic architecture of diseases and discovering potential new therapeutic targets.

Methods

The core of NETMAGE is its ability to generate interactive "disease-disease networks" (DDNs). In these networks, each node represents a phenotype (disease or trait), and an edge connects two nodes if they share a significant genetic association.

The NETMAGE tool itself is built upon the InteractiveVis framework, which uses sigma.js for network visualization. A key innovation is its flexible web interface that can automatically parse user-provided PheWAS summary data. This means researchers don't need to conform to a rigid data format; NETMAGE can adapt to various input structures. The tool allows users to:

  • Upload PheWAS results: Users can provide their own summarized PheWAS data.
  • Generate interactive networks: The tool creates dynamic visualizations of phenotype associations.
  • Search and filter networks: Users can explore the network by various attributes like phenotype name, associated SNPs, or network statistics.
  • View detailed node information: Selecting a node reveals associated phenotypes, SNPs, and other relevant data.

For generating the network files, the authors provide a Python script called prepDataForGephi.py. This script takes raw PheWAS data and can optionally filter it based on p-value, minor allele frequency, and case count. It then generates "edge maps" (representing connections between phenotypes) and "node maps" (describing the phenotypes themselves). These maps are then processed using the open-source network visualization software Gephi to create the final interactive network.

Datasets

The primary dataset used for demonstrating NETMAGE's capabilities is the UK Biobank (UKBB) PheWAS summary data. This dataset includes:

  • 400,000 Caucasian British samples.
  • 1403 phenotypes expressed in terms of PheCodes.
  • 28 million genetic variants.

The data was analyzed using the SAIGE R package to produce p-values for SNP-disease associations. For the demonstration, the data was filtered using specific criteria (p-value < 1x10⁻⁸, minor allele frequency > 0.01, and case count > 200) to create a manageable network of 260 unique phenotypes and 2,642 genetic associations.

Key Insights and Ideas

  • Bridging the visualization gap: NETMAGE directly addresses the need for accessible, interactive visualization of complex PheWAS results, moving beyond static representations.
  • Uncovering pleiotropy: By visualizing shared genetic associations between phenotypes, the tool facilitates the identification of potential pleiotropic effects, where a single genetic locus influences multiple traits. This is crucial for understanding disease mechanisms and identifying drug targets.
  • Flexibility in data input: The ability to parse various input formats for PheWAS data makes NETMAGE broadly applicable to different research datasets.
  • Integration with existing tools: The workflow leverages Gephi for network construction, allowing users to benefit from its advanced visualization and analysis capabilities.
  • Potential for personalized medicine: By revealing complex genetic relationships, NETMAGE can guide drug design and contribute to the development of personalized medicine approaches.

Key Results

The NETMAGE tool was successfully applied to the UK Biobank data, generating an interactive network with 260 phenotypes and 2,642 genetic associations. The authors highlight that their map correctly identified previously known instances of pleiotropy, such as associations with specific SNPs and multiple related conditions (e.g., rs544873 with pulmonary heart disease, phlebitis, hemorrhoids, etc.). The tool also demonstrated its ability to search and filter the network, allowing users to focus on specific genetic associations. The generated network is publicly available at https://hdpm.biomedinfolab.com/ukbb/.

Limitations and Future Work

The authors acknowledge several areas for future development:

  • Gene-gene networks: They plan to introduce functionality to create "gene-gene networks" that depict associations between genes based on their shared effects on phenotypes.
  • Network comparison: The ability to compare multiple networks generated from different datasets is a desired future feature.
  • Validation on additional datasets: Running NETMAGE on datasets like the Penn Medicine Biobank will help evaluate the accuracy of identified genetic associations and distinguish true pleiotropy from potential false positives.

The current implementation focuses on visualizing phenotype-phenotype associations based on shared genetic links. While powerful, the direct visualization of gene-phenotype or gene-gene relationships would further enhance its utility.

Summary updated

Abstract

Abstract Summary Given genetic associations from a PheWAS, a disease-disease network can be constructed where nodes represent phenotypes and edges represent shared genetic associations between phenotypes. To improve the accessibility of the visualization of shared genetic components across phenotypes, we developed the humaN-disEase phenoType MAp GEnerator (NETMAGE), a web-based tool that produces interactive phenotype network visualizations from summarized PheWAS results. Users can search the map by a variety of attributes, and they can select nodes to view information such as related phenotypes, associated SNPs, and other network statistics. As a test case, we constructed a network using UK BioBank PheWAS summary data. By examining the associations between phenotypes in our map, we can potentially identify novel instances of pleiotropy, where loci influence multiple phenotypic traits. Thus, our tool provides researchers with a means to identify prospective genetic targets for drug design, contributing to the exploration of personalized medicine. Availability and implementation Our service runs at https://hdpm.biomedinfolab.com . Source code can be downloaded at https://github.com/dokyoonkimlab/netmage . Contact [email protected] Supplementary information Supplementary data and user guide are available at Bioinformatics online.
BibTeX
@misc{sriram2020netmage,
  author = {Vivek Sriram and Manu Shivakumar and Sang‐Hyuk Jung and Lisa Bang and Anurag Verma and Seunggeun Lee and Eun Kyung Choe and Dokyoon Kim},
  title = {{NETMAGE: a humaN-disEase phenoType MAp GEnerator for the Visualization of PheWAS}},
  year = {2020},
  doi = {10.1101/2020.10.27.357103},
  howpublished = {bioRxiv},
}
Figure