I may not be what many people think of when they imagine a scientist. I don't wear a white coat, mix chemicals or take measurements, and I don't form hypotheses or run experiments. I make tools.

Specifically, I write computer programs to help other scientists in their work. My signature program helps researchers to find genes that influence the risk of disease. This isn't just an exercise in computer coding: my colleagues and I have combined a wealth of experience in statistics and population genetics to build sophisticated models of the way in which genes are inherited over many generations. When applied thoughtfully, these models can increase the value of existing data sets and hasten the process of discovery.

Although I love making scientific tools, perhaps as a result of my undergraduate experience in an engineering department, I have some trepidation about staying on this track as I prepare to apply for jobs in academic research. On the whole, prominent journals tend to value data over methods, making it difficult to publish unless one's shiny new algorithm can ride on the coat-tails of an interesting data set — and even then there may be disputes about authorship. Access to cutting-edge data is also crucial to building a good tool in the first place, because it's hard to model data you've never seen. Partnerships between 'wet' and 'dry' labs can alleviate this concern and provide mutual benefits, but such relationships can be hard to establish, especially when you're a young toolmaker.

It is also immensely time-consuming to create and maintain a good software tool. Converting a statistical idea into a working computer program is the easy part; the hard part is accounting for unexpected input, anomalous data sets and specialized extensions, all while keeping the program fast and easy to use. These mundane chores are frustrating because I don't feel like I'm 'doing science' during the many hours they consume, which I would rather spend hatching new ideas, running analyses or writing papers. A widely used software tool represents a significant contribution to the field, but I worry that the time it takes away from publishing will damage my chances of landing a faculty job.

Nevertheless, I relish my role. Rather than being chained to a single line of enquiry, I contribute to many research projects at the same time. On learning that I am a human geneticist, people typically ask which disease I work on, and it is gratifying to say “most of them”. My entire toolbox lives inside my laptop: instead of tending cell cultures on Saturdays or hovering over PCR machines under fluorescent lights, I can work wherever the sun is brightest and the coffee is thickest.

Toolmakers play an essential role in the scientific endeavour, and our importance will only grow as technological advances produce larger and more complex data sets. Still, sometimes I wonder: is tool-making a viable path to a career in academia?