Advances in science and computing technology have accelerated the development and dissemination of a wide range of big data platforms such as bioinformatics into the biomedical and life sciences environments. Bioinformatics brings the promise of enabling life scientists to easily and effectively access large and complex data sets in new ways, thus promoting scientific discoveries by for example generating, validating, and refining hypotheses based on in silico analysis (performed on computer). Meanwhile, life scientists still face challenges in working with big data sets such as difficulties in data extraction and analyses arising from distributed and heterogeneous databases, user interface inconsistencies and discrepancies in results. Moreover, the interdisciplinary nature of modern science adds to significant gaps in scientists' performance caused by limited proficiency levels with bioinformatics resources and a lack of common language across different disciplines.

Although developers of bioinformatics platforms are slowly beginning to move away from function-oriented software engineering approaches and towards to user-centered design approaches, they rarely consider users' value, and expectations that embrace different user contexts. Further, there is an absence of research that specifically aims to support the broad range of users from multiple fields of study, including 'wet' (lab-based) and dry' (computational) research communities.

Therefore, the ultimate goal of this research is to investigate life scientists' user experiences with knowledge resources and derive design implications for delivering consistent user experiences across different user classes in order to better support data-intensive research communities. To achieve this research goal, we used the theory of distributed cognition as a framework for representing the dynamic interactions among end users and knowledge resources within computer-supported and -mediated environments. To be specific, this research focused on how online bioinformatics resources can be improved in order to both mitigate performance differences among the diverse user classes and better support distributed cognitive activities in data-intensive interdisciplinary research environments. This research consists of three parts: (1) understanding user experience levels with current bioinformatics resources and key determinants to encourage distributed cognitive activities, especially knowledge networking, (2) gaining in-depth understanding of scientists' insight generation behavior and human performance associated with individual differences (i.e., research roles and cognitive styles), and (3) identifying in-context usefulness, and barriers to make better use of bioinformatics resources in real working research contexts and derive design considerations to satisfactorily support positive user experiences. To achieve our research goals, we used a mixed-methods research approach that combines both quantitative (Study 1 and 2) and qualitative (Study 3) methods.

First, as a baseline for subsequent studies, we conducted an empirical survey to examine 1) user experience levels with current bioinformatics resources, 2) important criteria to adequately support user requirements, 3) levels of knowledge networking (i.e., knowledge sharing and use) and relationship to users' larger set of distributed cognitive activities, and, 4) key barriers and enablers of knowledge networking. We collected responses from 179 scientists and our findings revealed that lack of integration, inconsistent results and user interfaces across bioinformatics resources, and perceived steep learning curves are current limitations to productive user experiences. Performance-related factors such as speed and responsiveness of resources and ease of use ranked relatively high as important criteria for bioinformatics resources. Our research also confirmed that source credibility, fear of getting scooped, and certain motivation factors (i.e., reciprocal benefit, reputation, and altruism) have an influence on scientists' intention to engage in distributed cognitive activities.

Second, we conducted a laboratory experiment with a sample of 16 scientists in the broad area of bench and application sciences. We elicited 1) behavior characteristics, 2) insight characteristics, 3) gaze characteristics, and 4) human errors in relation to individual differences (i.e., research roles such as bench and application scientists, cognitive styles such as field-independent and dependent people) to identify whether human performance gaps exist. Our results (1) confirmed significant differences with respect to insight generation behavior and human performance depending on research roles, and (2) identified some relationships between scientists' cognitive styles and human performance.

Third, we collected a rich set of qualitative data from 6 scientists using a longitudinal diary study and a focus group session. The specific objective of this study was to identify in-context usefulness and barriers to using knowledge resources in a real work context to subsequently derive focused design implications. For this work, we examined 1) the types of distributed cognitive activities participants performed, 2) the challenges and alternative actions they faced, 3) important criteria that influenced tasks, and 4) values to support distributed cognitive activities. Based on the empirical findings of this study, we suggest design considerations to support scientists' distributed cognitive activities from user experience perspectives.

Overall, this research provides insights and implications for user interface design in order to support data-intensive interdisciplinary communities. Given the importance of today's knowledge-based interdisciplinary society, our findings can also serve as an impetus for accelerating a collaborative culture of scientific discovery in online biomedical and life science research communities. The findings can contribute to the design of online bioinformatics resources to support diverse groups of professionals from different disciplinary backgrounds. Consequently, the implications of these findings can help user experience professionals and system developers working in biomedical and life sciences who seek ways to better support research communities from user experience perspectives.