compariSeq

By: Sean McKenna, Alex Bigelow, P. Samuel Quinan

Sequence logos are one of the most common methods to display biological sequences. Accurately comparing residue frequencies using sequence logos, however, is extremely difficult due to the combination and vertical stacking and scaling relative to information content. Using observations garnered from informal interviews with biologists of varying expertise, our proposed redesign, compariSeq, was specifically formulated to better support such comparison tasks.

Installation

This tool is a prototype written for the Processing library, and you will have to download the Processing application and open the code in compariSeq/compariSeq.pde and hit play to run the tool. It is provided as-is, and we have included the data from the redesign contest as an example of the required data input. You will likely want to replace this data with your own, which are the three text files in compariSeq/data. It will read these sequences and parse the data for visualization.

Please note that aspects such as screen size and how many items to display may be limited due to how the code was written. Once inside the tool, you can left or right click on the left to open up a detail view, and hovering over the provided legend will highlight the corresponding amino acid. You can also hit the key p at any time to output a PDF screenshot of the tool.

Prototype Overview

The main view of compariSeq shows the full sequence, allowing users to see general trends and locate interesting features. We vertically orient the sequences, allowing multiple sequences to be compared while simultaneously. At each location the bar length encodes conservation, the pie chart encodes relative frequencies of all amino acids, and the most common amino acids are also listed left to right.

We also provide a linked view that provides a direct comparison of frequency information across sequences for selected locations. Relative frequencies are encoded though the position along a common axis, while size simultaneously denotes conservation.

Design Justification

We identified the comparison of specific locations across sequences as a primary task, observing that domain knowledge, coupled with conservation and approximate frequencies, is enough to determine which locations are most interesting. As such, we made conservation the most salient feature in our design, encoding it in a bar chart to enable users to quickly cue into locations of interest. Once interesting locations have been identified, users came more about frequency comparison. The pie charts and detail views enable such comparisons while our overview also provides explicit access to 'consensus' through a vertical scan.

Acknowledgements

We would especially like to thank Heidi Schubert, Matthew Sdano, Venkatesh Rajamanikkam, and Miriah Meyer for the information and feedback they provided on our design.