Exploring Genetic Data Across Individuals: Design and Evaluation of a Novel Comparative Report Tool

Abstract

Background: The growth in the availability of personal genomic data to nonexperts poses multiple challenges to human-computer interaction research; data are highly sensitive, complex, and have health implications for individuals and families. However, there has been little research on how nonexpert users explore their genomic data.

Objective: We focus on how to support nonexperts in exploring and comparing their own personal genomic report with those of other people. We designed and evaluated CrossGenomics, a novel tool for comparing personal genetic reports, which enables exploration of shared and unshared genetic variants. Focusing on communicating comparative impact, rarity, and certainty, we evaluated alternative novel interactive prototypes.

Methods: We conducted 3 user studies. The first focuses on assessing the usability and understandability of a prototype that facilitates the comparison of reports from 2 family members. Following a design iteration, we studied how various prototypes support the comparison of genetic reports of a 4-person family. Finally, we evaluated the needs of early adopters-people who share their genetic reports publicly for comparing their genetic reports with that of others.

Results: In the first study, sunburst- and Venn-based comparisons of two genomes led to significantly higher domain comprehension, compared with the linear comparison and with the commonly used tabular format. However, results show gaps between objective and subjective comprehension, as sunburst users reported significantly lower perceived understanding and higher levels of confusion than the users of the tabular report. In the second study, users who were allowed to switch between the different comparison views presented higher comprehension levels, as well as more complex reasoning than users who were limited to a single comparison view. In the third study, 35% (17/49) reported learning something new from comparing their own data with another person's data. Users indicated that filtering and toggling between comparison views were the most useful features.

Conclusions: Our findings (1) highlight features and visualizations that show strengths in facilitating user comprehension of genomic data, (2) demonstrate the value of affording users the flexibility to examine the same report using multiple views, and (3) emphasize users' needs in comparison of genomic data. We conclude with design implications for engaging nonexperts with complex multidimensional genomic data.

Keywords: consumer health informatics; genomics.

Figure 1

Tabular report: the table is similar to the existing GET-Evidence report with each row representing a variant in one or both of the reports. Two columns were added to the table and check marks were used to denote the presence of a gene variant for each sibling.

Figure 2

Linear visualization: each rectangle represents a gene variant. Jamie’s variants are represented along the top, Alex’s variants are represented along the bottom, and their shared variants grouped to the left.

Figure 3

Sunburst visualization: each arc represents a gene variant. The inner circle represents Jamie’s variants and the outer circle represents Alex’s. Variants in both circles represent the ones Jamie and Alex have in common.

Figure 4

Venn visualization: displays a Venn diagram of gene variants. The bubbles on the left represent Jamie’s variants and bubbles on the right are Alex’s. Bubbles in the middle represent the variants Jamie and Alex have in common.

Figure 5

Tabular report: the table is similar to the existing GET-Evidence report, with each row representing a variant in any of the reports. Four columns were added to the table and a checkmark or a carrier indicator was used to denote the presence of a gene variant for each family member. We added a new search and filters bar.

Figure 6

Linear visualization: each rectangle represents a gene variant and each row represents a family member’s genome. Parent 1’s variants are represented at the top, followed by Parent 2’s variants, then Child 1’s variants, then Child 2’s variants at the bottom. The colored variants in each row represent the variants of one family member.

Figure 7

Sunburst visualization: each arc represents a gene variant, and each full circle represents a family member’s genome. The outer circle represents Parent 1’s variants, followed by Parent 2’s variants, then Child 1’s variants, and then Child 2’s variants in the inner circle. The colored variants in each circle represent the variants of one family member.

Figure 8

Comparison tab of CrossGenomics 2.1, where users can toggle between the four comparison visualizations using the buttons under “Change Views”.

Figure 9

Overview tab of CrossGenomics 2.1 in graph view, where users can view their own personal genomic data graphed by certainty of evidence and potential health effect in our single-user GenomiX visualization.

Figure 10

Overview tab of CrossGenomics 2.1 in category view, where users can view their own personal genomic data by health category in our single-user GenomiX visualization.