My research project at Brown University involved trying to dissect the inner-workings of human t-cells – the shepherds of our immune system. By mutating individual proteins inside the cell and measuring the effect on the thousands of other proteins around it, we hoped to discover something new about how our bodies fight diseases. In an certain aspect, my research was exciting because the technology required to generate this magnitude and this type of data was developed only years ago. On the other hand, this type of data analysis is often overwhelming because tools to work with this data effectively are still lacking. In addition to my role as a researcher in lab, I also helped develop a visual system to analyze and present our data.
When brainstorming on how to design graphics, I set out with a few questions in mind:
- What information will be necessary for the graphics be sufficiently self-explanatory?
- How can I reduce visual noise?
- How can I make the data more understandable?
- When publishing our results, I found that it was confusing to simply report tables of numbers. By showing numbers as colors, readers could visually identify a single profile of a protein and its activity pattern.
In addition, by overlaying our data over existing solved pathway structures, it helped readers to orientate themselves as well as make sense of our data. I was unsatisfied with how existing pathway structures were designed. Some were unnecessarily complex (by trying to show too many unimportant details) while others were too simply and left out crucial details. I came up with my own system of pathway diagrams that focused on emphasizing important points and reducing the “spaghetti factor” (confusion caused by too many intersecting lines).
Because our research was not just about producing findings, but also to describe a novel method of conducting experiments, it was important for us to clearly lay out each step of our experiment – and to break down how different steps relate to different components of the experiment (SILAC quantitation vs label-free quantitation).