Communications of the ACM
Home/Magazine Archive/November 2004 (Vol. 47, No. 11)/Beyond Power: Making Bioinformatics Tools User-Centered/Full Text
Bioinformatics: transforming biomedical research and medical care
Beyond Power: Making Bioinformatics Tools User-Centered
Providers of bioinformatics data face a conundrum in the way Web-accessible services are viewed and used. Biologists might recall the torturous process of obtaining genetic data prior to the advent of the Web. But students and scientists new to the field find that Web-based bioinformatics tools lack the sophistication of the interfaces they've grown up with in computer games and personal productivity tools (such as word processing and spreadsheets). The attitudes of these potential new users will thus continue to limit the rate of adoption and use of Web-based bioinformatics tools. How bioinformatics contributes to human health care will be dramatically influenced by the degree to which the tools are adopted and used as a routine aspect of biomedical research.
User-centered design methodologies may be used by providers and developers of Web-based bioinformatics systems to build services that are more usable by novices and more powerful for experts. Attention to this aspect of bioinformatics will broaden the use of genomic information in medical research and practice while enabling new insight into biological processes. Here, we explore the core user-centered design concepts that we, as part of the Human-Centered Software Engineering (HCSE) Group at Concordia University, are applying to our studies of bioinformatics Web sites, in order to help improve their utility to the biomedical research community [3].
Increased usability has been linked to decreased training costs and time, as well as to easing the transition to new versions of existing systems, improving human performance and productivity, ensuring better quality of work, and minimizing the risk of user error in data entry [5]. Usability is critical to the future of bioinformatics tools because the biomedical research community involves high-cost scientific personnel, laboratory experimentation to generate data, and bioanalytical techniques to analyze that data.
Designing a user interface (UI) for bioinformatics tools to make them better able to convey information or organize a body of data, information, or knowledge is a demanding proposition. The complexity and diversity surrounding usability of bioinformatics systems involves diversity in the tools themselves, as well as in their users. A variety of tools is available; for example, the National Center for Biotechnology Information (NCBI) Web site (www.ncbi.nlm.nih.gov) is an information portal with access to specialized tools (such as for sequence alignment and molecular visualization). Users accessing and interacting with them pursue goals ranging from simple information gathering (such as article searching in Pub Med) to specific molecular biology problems (such as with the Basic Local Alignment Search Tool, or BLAST).
The first step in applying user-centered design methodologies and principles [6, 7] is to gain a better understanding of biologists' personal behavior, experience, tasks, and work context. Empirical studies involving all aspects of biologists' experience and interaction with popular tools are therefore necessary for determining how the tools are perceived, learned, and mastered. Only in this way can a usability expert perform a task analysis and scenario generation, followed by low-fidelity prototyping and rough usability studies, ultimately yielding a better bioinformatics tool. Table
Besides our group, other HCI teams around the world are engaged in related areas: bioinformatics visualization and usability (Ben Shneiderman at the University of Maryland); zoomable interfaces for genome navigation and comparing multiple genomes and proteomes (William Ribarsky at Georgia Tech); task-based and user-centered protocols for bioinformatics information search and retrieval (Joan Bartlett at McGill University); and analysis and classification of tasks in bioinformatics (Robert Stevens at University of Manchester).
Personae and Pattern-Oriented Design
In 20022003, as part of the BioUse Project, the HCSE Group's usability lab conducted ethnographic interviews and usability studies with 16 biologists and three UI experts using the NCBI site. We used the information to understand the problem domain (bioinformatics) and user goals and expectations for the related tools. We interviewed a total of 19 participants from North American and European research groups, including biologists, medical practitioners, and computer scientists (see the table here).
The complexity and diversity surrounding the usability of bioinformatics systems involves diversity in the tools themselves, as well as in their users.
We divided these users into two personae: novices and experts. The concept of personae helps place a particular user of a particular application at the center of the design process [1]. Personae are based on user characteristics (such as work habits, experience, and expertise). Personae permit UI designers to categorize users into a small number of groups with consistent expectations, skills, and preferences. Usability studies have found considerable differences in the experience reported by expert vs. novice personae. For example, the percentage of users satisfied with the NCBI site was uniformly greater among experts than among novices when tested against nine different usability heuristics (see Figure 1). Our ethnographic interviews confirmed that novice users find the learning curve for such sites steeper than they would prefer.
The software design concept of pattern-oriented design (POD) may be used by software engineers and usability experts to create Web sites that customize the solution of common problems for different personae. A pattern is a proven solution to a common user problem that occurs in many different contexts [9]. For example, "advanced search" can be described as a pattern with certain attributes, including allowing expert users to search with extended-term matching, along with scoping and output options. A "wizard" is another pattern that allows novice users to achieve a particular goal when several intermediate decisions must be made first. We used POD to redesign the NCBI site to support both novice and expert users, receiving positive feedback (in the form of the ethnographic interviews) from the original participants. POD supplements empirical studies with personae by associating patterns to personae and using them as a design tool [4]. For example, a handful of patterns can be associated to the personae and related empirical studies in five aspects of Web-based applications:
- Information architecture. Cognitive organization of content;
- Screen layout. Organization of content and actions on working surfaces;
- Navigation. Interaction mechanisms;
- Information visualization. Visual representations and metaphors for grouping information in cognitively accessible segments; and
- Communication and interoperability. Code modules within an application.
By taking advantage of existing repositories of patterns, designers can build Web-based tools that provide multiple views ofor portals intothe core functionality of a given bioinformatics Web site, thus optimizing it for multiple personae. In addition, by understanding biologists' usage patterns, usability and software engineers can organize tools more intuitively using a task-based approach to reflect the steps involved in the bioanalytical process.
Since human-centered tools concentrate on user behavior, anyone designing bioinformatics tools must first analyze and understand biologists' experience to be able to build models of their behavior.
The HCSE Group is building the BioUse Web portal as an easy-to-use interface customizable for a variety of user personae. A portal for each type of personae would serve as a front-end to a number of tools, including BLAST for homology searching and ClustalW for multiple sequence alignment. The portal groups tools on the screen in ways a user would find intuitive, using a task- and process-based approach. Over time, as users become increasingly familiar with the application, the portal adds shortcuts. It also includes personalization features in terms of tools and links that are of greatest interest to individual users depending on their personal domains; for example, pharmacologists clearly have different interests from microbiologists. The BioUse portal prototype is being implemented through the POD technique using reusable Web components [2] with XML technology. The novice view of the portal guides users through the bioanalytical process, describing each task in turn (see Figure 2).
Incorporating the concepts of personae and pattern use, the portal helps usability experts and software engineers put the sometimes diverse needs of biologists at the center of the design process. They create model personae, using them to evaluate existing UIs, then apply POD to improve Web interfaces for biomedical applications. These methods may improve the effectiveness of tools for existing users and increase the rate at which they (despite their inherent complexity) are adopted by even novice biomedical researchers.
Conclusion
We intend to explore further all aspects of human experience and interaction with these tools and Web sites, focusing on ease-of-use and biologists' research needs, in order to improve functionality. Since human-centered tools concentrate on user behavior, anyone designing bioinformatics tools must first analyze and understand biologists' experience to be able to build models of their behavior. Only with such models are they prepared to take on bioinformatics tool design and release the tools for public consumption.
Although the amount of bioinformatics information and tools is increasing dramatically, BioUse is one of the few such projects in which users themselves are indeed the research focus. The complexity of the bioinformatics domain represents a challenge to biologists and bioinformatics tool designers alike, so close interaction is a must. Tools designed specifically for the biomedical user community are likely to be adopted only if they concentrate on and adapt to the people who are ultimately expected to use them.
References
1. Cooper, A. The Inmates are Running the Asylum. SAMS Publishing, Indianapolis, IN, 1999; www.cooper.com/content/insights/cooper_books.asp.
2. Dotalis R&D Group. Montreal, Canada; www.dotalis.com/.
3. Human-Centered Software Engineering Group. Bio-Use Project. Concordia University, Montreal, Quebec, Canada; hci.cs.concordia.ca/ www/hcse/projects/bio-use/.
4. Javahery, H. and Seffah, A. A model for usability pattern-oriented design. In Proceedings of the First International Workshop on Task Models and Diagrams for User Interface Design, C. Pribeanu and J. Vanderdonckt, Eds. (Bucharest, Romania, July 1819, 2002), 104110.
5. Mayhew, D. Usability Engineering Lifecycle. Morgan Kauffman, San Francisco, 1999.
6. Nielsen, J. Heuristic evaluation. In Usability Inspection Methods, J. Nielsen and R. Mack, Eds. John Wiley & Sons, Inc., New York, 1994.
7. Seffah, A. Learning the ropes: Human-centered design skills and patterns for software engineers' education. Interact. 10, 5 (Sept.Oct. 2003), 3645.
8. UsablityNet. ISO 13407. Human-centered design processes for interactive systems (1999); www.usabilitynet.org/tools/13407stds.htm.
9. Welie, M. and Traetteberg, H. Interaction patterns in user interfaces. In Proceedings of the Seventh Pattern Languages of Programs Conference (Monticello, IL, Aug. 1316, 2000); jerry.cs.uiuc.edu/~plop/plop2k/ proceedings/Welie/Welie.pdf.
Footnotes
This work is supported by the National Sciences and Engineering Research Council of Canada, le Fonds Québécois de la Recherche sur la Nature et les Technologies, and the Faculty of Engineering of the Concordia Research Chair programs.
Figures
Figure 1. Satisfaction ratings of novice and expert users of the NCBI Web site.
Figure 2. Novice user view of the BioUSE portal.
Figure. Intuitively following the process of analyzing a gene sequence from the biologist's point of view, bioinformatics tool designers create process-sensitive tools that are more usable and more likely to be used. Rozita Naghshin, Human-Centered Software Engineering Group, Concordia University, Montreal.
Figure. Visualization of the Visible Mouse based on data obtained from the Center for In Vivo Microscopy at Duke University using the Pittsburgh Supercomputing Center's Volume Browser. Art Wetzel, Stu Pomerantz, De'mian Nave, and David Deerfield at the Pittsburgh Supercomputing Center, Carnegie Mellon University.
©2004 ACM 0001-0782/04/1100 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2004 ACM, Inc.
No entries found