Dissecting Complexity: The Hidden Impact of Application Parameters on Bioinformatics Research

Biology is a quest; an ongoing inquiry about the nature of life. How do the different forms of life interact? What makes up an ecosystem? How does a tiny bacterium work? To answer these questions biologists turn increasingly to sophisticated computational tools. Many of these tools are highly configurable, allowing customization in support of a wide range of uses. For example, algorithms can be tuned for precision, efficiency, type of inquiry, or for specific categories of organisms or their component subsystems. Ideally, configurability provides useful flexibility. However, the complex landscape of configurability may be fraught with pitfalls. This paper examines that landscape in bioinformatics tools. We propose a methodology, SOMATA, to facilitate systematic exploration of the vast choice of application parameters, and apply it to three different tools on a range of scientific inquires. We further argue that the tools themselves are complex ecosystems. If biologists explore these, ask questions, and experiment just as they do with their biological counterparts, they will benefit by both finding improved solutions to their problems as well as increasing repeatability and transparency. We end with a call to the community for an increase in shared responsibility and communication between tool developers and the biologists that use them in the context of complex system decomposition.