Ten simple rules for developing visualization tools in genomics

• Published in: PLoS computational biology Vol. 18; no. 11; p. e1010622
• Main Authors: Durant, Eloi, Rouard, Mathieu, Ganko, Eric W, Muller, Cedric, Cleary, Alan M, Farmer, Andrew D, Conte, Matthieu, Sabot, Francois
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 01.11.2022; PLOS; Public Library of Science (PLoS)
• Subjects: Binding sites; Biology and Life Sciences; Cognitive ability; Computer and Information Sciences; Coronaviruses; Datasets; Design; Education; End users; Engineering and Technology; Genomics; Humanities and Social Sciences; Life cycles; Life Sciences; Medicine and Health Sciences; Methods; Quantitative Methods; Social Sciences; Software; Visual perception; Visualization; France
• ISSN: 1553-7358; 1553-734X; 1553-7358
• DOI: 10.1371/journal.pcbi.1010622

For better reading comfort, we organized the rules chronologically depending on when they would matter the most during a visualization tool’s life cycle, starting with the design (Rules 1 to 5) then development (Rules 5 to 8-ish) phases until it is shared with the rest of the world. Front-line analysts can help you to properly define the tool tasks (i.e., both high- and low-level tasks that your tool should achieve) and provide test data as well as valuable feedback during both the design and development phases. There are many good ways to engage with your end users, including face-to-face interviews, surveys, design sprints, and even hands-on sessions that can often reveal edge cases and sometimes bugs that would have been hard to find on your own—beta-testing at its finest. [...]there are fields dedicated to datavis: understanding how they are perceived by the human brain (visual perception and cognitive vision science), improving how they can be used with machines (Human Computer Interaction), and growing communities of datavis designers, full of people who could help build a visualization tool. Along with the genomic sequences, you may also have access to additional data types, such as Hi-C, epigenomic signatures, or detections of transcription factor binding sites, all of which can blur the respective message in all-in-one visualizations.