Potential Perils of Biological Sequence Visualization Using Sequence Logo

Abstract

Sequence motif's characteristics are commonly visualized by using a sequence logo. This paper describes a user study aimed at evaluating the effectiveness of sequence logo as evaluation metric for motif prediction tools. We also investigate the nature of confirmation biases in using sequence logos in result reporting in publications. While sequence logos have been widely used for visualizing sequence motifs in the past 20 years, no study has reported its effectiveness and possible misuses in decision making. We conducted a paper-and-pencil test to determine the effectiveness of sequence logos in some of their common usages. A survey study was also performed to investigate sequence logos' learn ability. We found that there are great mismatches between users' perception and actual quality of motifs when sequence logos were used as an evaluation metric. Therefore, evaluation of motif prediction tools based on sequence logos has to be interpreted cautiously. Our result also suggests that there are still room for improvements in the current sequence logo's layout design.