Beyond GWAS: Investigations of Accessory Methodologies to Improve Biological Interpretability

Open Access
- Author:
- Van Kampen, James
- Graduate Program:
- Neuroscience
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- December 10, 2020
- Committee Members:
- Dajiang Liu, Dissertation Advisor/Co-Advisor
Dajiang Liu, Committee Chair/Co-Chair
Andras Hajnal, Committee Member
James Riley Broach, Outside Member
Kent Eugene Vrana, Committee Member
Alistair J Barber, Program Head/Chair - Keywords:
- Addiction
Smoking
Genetics
GWAS - Abstract:
- As computational capacity and database size and complexity increase, genome wide association studies (GWAS) increase in power, allowing the identification of more potentially linked genome variants. While these methods are powerful on their own, they may be even more useful in the future in combination with other computational methods. Unfortunately, many of the identified variants are common and non-coding. It is possible that these variants are in linkage disequilibrium (LD) with rare variants. Modified computational methods that increase apparent sample size (meta-analysis), account for multiethnic backgrounds, or add in electronic health record or other phenotypic data, may provide further details as to the associations. Here, we report that integration of MRI GWAS data using TESLA to account for multiethnic sampling allows for the identification of novel significant genes in the context of smoking phenotypes, including age of initiation, smoking cessation, cigarettes per day, and smoking initiation. This technique allows us to identify genes that associate with the phenotype and with specific brain regions of interest. GWAS data-handling methods may also be useful in other areas, such as the use of pathway analysis to identify potential mechanisms of sensitivity and resistance to natural product derivatives with novel anticancer therapeutic profiles. For instance, pathway analysis of RNA sequencing data of cancer cells treated with schweinfurthin analogs revealed a differential expression of genes associated with the Sonic Hedgehog pathway in sensitive glioma (SF-295) and resistant lung (A549) cell lines. Although these computational methods are powerful, it is still important to validate findings via biological assays whenever possible. The identification of sonic hedgehog as a potentially important pathway was verified by inhibition of this pathway in the resistant cell line, second this was able to increase cell line sensitivity. Such findings suggest that co-treatment with Hedgehog pathway inhibitors may overcome cellular resistance to the iv anticancer effects of this class of natural compounds and derivatives, potentially increasing the usefulness of these compounds as future anticancer therapeutic agents. Similarly, many computational genomic studies have identified potential dysregulation of inflammation pathways and the hypothalamic-pituitary-adrenal (HPA) axis in bipolar disorder (BD), but biological assays were necessary to define the role of these potential alterations in the diurnal cortisol cycle, which regulates sleep and focus. Nighttime cortisol levels were significantly higher in BD patients, which may account for the well-documented sleep disturbances which patients commonly suffer. Such findings suggest that targeting cortisol patterns may be able to improve BD patient sleep and/or other outcomes. Clearly, computational genomic methods are a mainstay of modern research, but it is critical to continue to expand the area through the combination of data sets, new modeling methods, and biological validation. The present studies here investigate smoking phenotypes and common comorbid or risk diseases associated with smoking, bipolar disorder and cancer.