Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Kenneth Krauter

Second Advisor

Scott Vrieze

Third Advisor

Marissa Ehringer

Fourth Advisor

William Old

Fifth Advisor

Zoe Donaldson


Substance abuse is one of the leading causes of morbidity and mortality in both the developing and the developed worlds. For example, approximately 88,000 people die each year from alcohol- related causes and the annual cost to society of alcohol misuse is estimated to be $249,000,000,000. Converging lines of evidence indicate that these behaviors are substantially heritable: twin and adoption studies have found significant genetic effects for initiation, intensity of use, dependence, and abuse for alcohol, tobacco, marijuana, and other drugs. Genome-wide association studies (GWAS) and candidate gene studies have found a number of robust associations between genetic variants and substance use and dependence. Twin studies have found evidence for common genetic liability across drugs and for distinct genetic influences on substance use initiation and on quantity of use and substance dependence after initiation.

Chapter 2 of this thesis describes a rare variant GWAS meta-analysis focused primarily on the role of exonic variants in alcohol and smoking behavior. Across 17 contributing studies, most using the Exome Chip, I assembled a total sample size of between 70,847 and 164,142 individuals for five standard phenotypes: cigarettes per day, smoking initiation, pack years, age of smoking initiation, and drinks per week. In this meta-analysis, I performed single variant tests, gene-based burden tests, and tests conditioned on the effects of common variants. I replicated a number of known associations but failed to find any reproducible novel associations. A modest portion of phenotypic variance (1.7-3.6%) was accounted for by all genotyped rare variants. In summary, if rare variants with large effect sizes exist for these traits, they must be substantially more rare than the modestly rare variants genotyped on the Exome Chip. It follows that large sequenced samples will be required to detect their effects.

Chapter 3 of this thesis describes a twin study of adolescent substance use development which used smartphone applications and location tracking to measure twins’ behavior and environmental exposures. Adolescence is a sensitive period for substance use. Individuals who initiate substance use in early adolescence are at higher risk for dependence diagnoses in adulthood than individuals who initiate later in adolescence. Excessive adolescent substance use is also associated with increased risk for accidental and intentional injuries. Genetic and environmental explanations for adolescent substance use have been advanced but few studies have examined specific environmental hypotheses while accounting for genetic confounding. In this chapter, I show that substance use behavior and related variables can be measured accurately and at high frequency by automated remote assessment and monitoring mediated through the participant’s smartphone. I found that adolescent substance use and change in use is heritable, including e-cigarette use, a novel result. I used the participants’ location data to measure the fraction of time they spent at school during the school day and at home at night, measures of delinquent behavior. These variables were not associated with substance use, contradicting previous results. I also found that the physical distance between twins in a twin pair increased with age and increased more quickly for dizygotic twins than for monozygotic twins, a violation of the equal environment assumption of the classic twin model and an indication that location is heritable. In conclusion, digital phenotyping methods can be used to obtain high quality, longitudinal data in a scalable fashion.

Available for download on Sunday, October 10, 2021