Dataflow programming for the analysis of molecular dynamics with AViS, an analysis and visualization software application

• Published in: PloS one Vol. 15; no. 4; p. e0231714
• Main Authors: Pua, Kai, Yuhara, Daisuke, Ayuba, Sho, Yasuoka, Kenji
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 21.04.2020; Public Library of Science (PLoS)
• Subjects: Algorithms; C plus plus; Computer and Information Sciences; Computer programs; Computer simulation; Computers; Connectors; Data visualization; Depth perception; Engineering and Technology; Extensibility; Flexibility; Hydrogen bonds; Integrated software; Mechanical engineering; Molecular dynamics; Molecular structure; Nodes; Physical Sciences; Research and Analysis Methods; Software; Source code; Space perception; User interface; Variables; Visualization; Workflow; Japan
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0231714

The study of molecular dynamics simulations is largely facilitated by analysis and visualization toolsets. However, these toolsets are often designed for specific use cases and those only, while scripting extensions to such toolsets is often exceedingly complicated. To overcome this problem, we designed a software application called AViS which focuses on the extensibility of analysis. By utilizing the dataflow programming (DFP) paradigm, algorithms can be defined by execution graphs, and arbitrary data can be transferred between nodes using visual connectors. Extension nodes can be implemented in either Python, C++, and Fortran, and combined in the same algorithm. AViS offers a comprehensive collection of nodes for sophisticated visualization state modifications, thus greatly simplifying the rules for writing extensions. Input files can also be read from the server automatically, and data is fetched automatically to improve memory usage. In addition, the visualization system of AViS uses physically-based rendering techniques, improving the 3D perception of molecular structures for interactive visualization. By performing two case studies on complex molecular systems, we show that the DFP workflow offers a much higher level of flexibility and extensibility when compared to legacy workflows. The software source code and binaries for Windows, MacOS, and Linux are freely available at https://avis-md.github.io/.