PEAX: Interactive Visual Analysis and Exploration of Complex Clinical Phenotype and Gene Expression Association

Michael A. Hinterberg, David P. Kao, Michael R. Bristow, Lawrence E. Hunter, J. David Port, Carsten Görg

Abstract

In the era of personalized medicine, researchers face the challenge of integrating deep clinical phenotypes with massive molecular profiles. PEAX (Phenotype-Expression Association eXplorer) is an open-source visual analytics tool designed to bridge the gap between clinical expertise and high-dimensional genomic data. By integrating interactive decision trees with a high-performance statistical backend, PEAX allows researchers to define complex patient sub-groups and instantly visualize their association with gene expression. Validation using the BORG study demonstrates that PEAX can reduce a day-long manual analysis into a 30-minute exploratory session, facilitating rapid hypothesis generation and biological discovery.

Introduction

The workflow to connect clinical response data with genomic profiles is often inefficient. Typically, a clinical expert defines a phenotype using one set of tools and then relies on a statistician to perform the numerical analysis in a separate environment. This cyclical process is slow and can hinder the momentum of discovery. PEAX addresses this by putting statistical power directly into the hands of clinicians through a responsive, unified visual interface that supports real-time exploration of high-dimensional data.

Methodology

PEAX is built on a modern web stack optimized for data science. The frontend utilizes D3.js for interactive, collapsible decision trees and Shiny for the web interface. For phenotype modeling, the tool employs binary decision trees, which are intuitive for clinicians and allow for precise control over patient stratification compared to "black-box" machine learning models.

The statistical backend is powered by R. To maintain UI responsiveness, standard Analysis of Variance (ANOVA) functions were optimized. By decomposing mathematical operations into parallelized colMeans and colSums using the parallel library, the system can test thousands of genes across an 8-core system in under 30 seconds.

Visual Interface and Design

A key feature of PEAX is the use of "scented widgets"—sliders for phenotype thresholds that include embedded histograms. These widgets show the data distribution before a user makes a selection, providing immediate context for decision-making.

The PEAX interface showing side-by-side decision trees (C), scented widgets (B), and statistical results (D, E, F).

The tool allows for the side-by-side comparison of two different phenotype models. This juxtaposition enables researchers to observe how adding a single variable, such as a specific Single Nucleotide Polymorphism (SNP), alters the gene expression landscape across the cohort. Furthermore, the thickness of the lines in the decision tree varies based on the number of patients in a sub-group, providing immediate visual feedback on the robustness of the sample size.

Results and Validation

The authors validated PEAX using the BORG study, which examines the effect of $\beta$-blockers on heart failure patients. The dataset included longitudinal clinical data, myocardial gene expression (~34,000 mRNAs and ~7,800 miRNAs), and genotype data (SNPs).

Efficiency Gains

A cardiology expert was able to replicate previous research and explore new hypotheses in under 30 minutes—a task that previously required a full day of manual scripting. The reduction in lag from one day to 30 seconds was sufficient to maintain the "flow state" necessary for effective hypothesis generation.

Biological Discovery

Using PEAX, researchers identified a significant gene association ($p < 0.0005$) that only became apparent when analyzing the interaction between a specific $\beta$1-adrenergic receptor SNP and the drug treatment group. This highlights the tool's ability to uncover complex relationships that might be missed in standard linear analyses.

Discussion and Limitations

While PEAX significantly lowers the barrier to data exploration, it introduces a risk of "data dredging" or false discovery due to the high number of combinations tested. Future iterations should include more robust False Discovery Rate (FDR) adjustments. Additionally, while the tool identifies correlations between miRNA and mRNA, it does not yet integrate biological regulatory databases to explain the mechanisms behind these interactions.

Conclusion

PEAX demonstrates that interactive visual analytics can transform genomic research by empowering clinicians to lead the data exploration process. By combining intuitive modeling with high-performance computing, the tool facilitates a more agile and insightful approach to personalized medicine.

Summary updated

Abstract

Increasing availability of high-dimensional clinical data, which improves the ability to define more specific phenotypes, as well as molecular data, which can elucidate disease mechanisms, is a driving force and at the same time a major challenge for translational and personalized medicine. Successful research in this field requires an approach that ties together specific disease and health expertise with understanding of molecular data through statistical methods. We present PEAX (Phenotype-Expression Association eXplorer), built upon open-source software, which integrates visual phenotype model definition with statistical testing of expression data presented concurrently in a web-browser. The integration of data and analysis tasks in a single tool allows clinical domain experts to obtain new insights directly through exploration of relationships between multivariate phenotype models and gene expression data, showing the effects of model definition and modification while also exploiting potential meaningful associations between phenotype and miRNA-mRNA regulatory relationships. We combine the web visualization capabilities of Shiny and D3 with the power and speed of R for backend statistical analysis, in order to abstract the scripting required for repetitive analysis of sub-phenotype association. We describe the motivation for PEAX, demonstrate its utility through a use case involving heart failure research, and discuss computational challenges and observations. We show that our visual web-based representations are well-suited for rapid exploration of phenotype and gene expression association, facilitating insight and discovery by domain experts.
BibTeX
@inproceedings{hinterberg2015peax,
  author = {Michael A. Hinterberg and David P. Kao and Michael R. Bristow and Lawrence E. Hunter and J. David Port and Carsten Görg},
  title = {{PEAX: Interactive Visual Analysis and Exploration of Complex Clinical Phenotype and Gene Expression Association}},
  booktitle = {Pacific Symposium on Biocomputing},
  year = {2015},
  url = {https://openreview.net/forum?id=kIY8mmpltR},
}
Figure