Designing visualization dashboards for aligning experimental and analytical results in systems genetics addiction research

Studies in mice can help us make sense of human disease, due to genetic orthologies (overlap). In studying mice, we can formulate and test hypotheses quickly, and have experimental controls not afforded by human subjects research.
We are just beginning to understand the major role that genes and microbes play in determining traits, including behaviors -- in mice and in humans.
Did you know? The gut provides ~95% of humans’ total body serotonin, and 50% of the body’s dopamine is stored in the gut. That’s why the gut can be known as your “second brain”!
Certain experimental data can tell us a lot about this complex relationship. In the context of addiction research, assays (tests) might include traits such as novelty-seeking behavior in new environments; personal hygiene habits; activity levels; and specific characteristics of drug use, such as self-administration, withdrawal responses, and overdose. Drugs under study include cocaine, morphine, nicotine, and alcohol. Unsurprisingly, mice, just like humans, exhibit a wide range of diversity in these measures. The overdose hazards of certain drugs might be 4x higher for certain genetic groups. Microbes that lessen withdrawal symptoms might not be present in certain genetic groups. We therefore shouldn’t treat substance use disorders (SUDs) with a one-size-fits-all policy. The challenge for making sense of this data is that it’s high-dimensional.
Each mouse has a lot of data associated with it - and there are a lot of mice.
For each mouse, there is microbe abundance data. For some, there are multiple abundance values, from different tissue samples.
Additionally, there is whole genome sequence (WGS) data for all the mice, as well as metadata such as their sex and pedigree.
Then there is trait data. Because there are multiple experiments (e.g. novelty-seeking behaviors, drug self-administration) there are several of these data sets too.
Luckily, statistical measures can help slice and dice the data, while visualization can provide dynamic, interactive views, with “details on demand”. The value of these tools goes beyond systems genetics, though. Many of the data transformations required for these visualizations could have broad utility for other challenges in high-dimensional information visualization.
An overview of the motivation for the research.
We introduce three data visualization projects related to systems genetics and addiction research. Historically, the challenge of biology research has been one of data collection. However, as measurement becomes easier, the challenge shifts towards data sensemaking. Here, researchers are asking questions about the relationships between genes, traits, and the microbiome, using visualization for data exploration. This work has implications for visualization of high-dimensional data in other domains, as it shows how statistical methods can support visualization filtering, aggregation, and information hierarchies to explore dense data.
PDF | Preprint | BibTeX
Robyn Ball
Jason Bubier
Elissa Chesler

Khoury Vis Lab — Northeastern University
* West Village H, Room 302, 440 Huntington Ave, Boston, MA 02115, USA
* 100 Fore Street, Portland, ME 04101, USA
* Carnegie Hall, 201, 5000 MacArthur Blvd, Oakland, CA 94613, USA