SCCUR 2017 Abstract (Eileen Choe)

Abstract

GRNsight v2: a web application for visualizing models of gene regulatory networks

GRNsight is a web application and service for visualizing small- to medium-scale gene regulatory networks (GRNs). A GRN consists of genes, transcription factors, and the regulatory connections between them which govern the level of expression of mRNA and protein from genes. GRNsight produces weighted or unweighted network graphs by representing genes as nodes and regulatory connections as edges, with colors, end markers, and thicknesses corresponding to the sign and magnitude of activation or repression. After GRNsight v1’s release, feedback from users motivated improvements and new features. In GRNsight v2, the graph bounding box is separated from the viewport and expands as far as required for the graph to converge to a steady state, allowing the force graph parameters to be fully applied. The ability to zoom and scroll the graph within the viewport was also added. New analysis features include options for showing edge weight values, customizing the edge thickness normalization factor, and setting the threshold for coloring the edges gray. The addition of features in GRNsight v2 increased the number of possible combinations of functions a user can perform on GRNsight. Thus, we developed an automatic client-side test generator which produces a systematic testing manual. On the server side, syntactic tests for malformed Simple Interaction Format (.sif) files were written, which report end user-targeted errors if an incorrectly formatted SIF file is imported into GRNsight. Future work includes developing a framework to visualize the gene expression over time through coloring the nodes in the network graph. The node coloring feature coupled with the visual and analysis features in GRNsight v2 will provide a more complete visual representation of the GRN, allowing GRNsight users to better derive insights from them. The GRNsight application http://dondi.github.io/GRNsight/ and code https://github.com/dondi/GRNsight are available under the open source BSD license.

(295 words)

Keywords: Scientific Visualization, Software Engineering, Bioinformatics

Research Results

According to Ernest Boyer’s forms of scholarship (Scholarship Reconsidered, 1990), this project involves the scholarship of integration and application. As such, its primary research results revolve around the development and continuing evolution of an application for use by bioinformatics colleagues, following software development best practices. We have successfully built upon GRNsight in v2 to expand its visualization and layout capabilities, giving the user more control over the visual display of the network graph. Based on feedback from initial usability testing by the GRNmap modeling group and undergraduates in the Biomathematical Modeling course at Loyola Marymount University, GRNsight's visualization features enable the interpretation of weight values more easily than one could from the numerical data alone. The normalization factor and grey edge threshold features in GRNsight v2 facilitate comparison of multiple graphs so that the thicknesses of the edges will be on the same scale and allows the user to gauge the importance of particular regulatory relationships by "graying out" weak connections. In v2, GRNsight’s testing framework has been expanded to consist of over 230 unit tests for quality control with 78% statement coverage, adhering to test-driven development software engineering best practices and improving on the 160-test, 73% statement coverage of GRNsight v1. The complete set of research results can be found at https://github.com/dondi/GRNsight.