Bio-data visualisation: Tree Clustering

June 2018 – September 2018

Main tasks: Developing web tool, conducting user testing

Role: Research assistant

Company: Aalto University | HICT

Problem

Visualising data can be extremely helpful in understanding it, but current popular graphs are often not enough for complex data systems. Making data visualisations more understandable and usable was one of the topics of interest of the User Interfaces group at Aalto University. The project I got involved in aimed to increase the ease of clustering families of bacterias with phylogenetic trees. This could help eg. biologists, who research cancer-related issues

Actions

My responsibilities were two-fold. First, I focused on improving the web visualisation tool and preparing it for user testing. For that, I implemented several modifications in the tool with JavaScript (JS) and the D3 visualisation library. Later, I wrote a script in python for automatic control of the tool. For testing the tool, it was important for us to understand how long a person looks at different parts of the screen. Therefore, I also wrote the code in C# that gathered eye-sigh data from the eye-tracker and combined it with the information on the screen.

The next responsibility was the user testing itself. I started with writing the qualitative questionnaire and the needed consent and information forms. Then, I prepared the lab for good experiment conditions and started inviting participants. During the experiment, I collected both qualitative data, via short interviews with the participant, as well as quantitate data from the tool and eye-tracking device. After collecting and cleaning the data, I made an initial analysis with python.

The user testing setup. User was marking the bacteria families on the screen, while their eyesight was measured with the eye-tracking tool.

Results

Our main goal was to check if our tool is increasing the easiness of the perception of the bacteria families on the phylogenetic trees. Thanks to my work on preparing the tool, data collecting scripts and experiment itself, we managed to receive valuable data from around 20 participants who took part in the testing sessions. Results of my work were much appreciated by the research team and used for the last improvements of the tool.