Project Overview
Computer Science and Mathematics Meet Biology
Department(s)
Computer Science
Mathematics
Biology
Abstract
My lab's research focuses on understanding biological systems and human disorders through mathematical modeling, machine learning, network analysis, and computer programming. Since 2010, I've supervised over 80 research students from a variety of disciplines (including computer science, mathematics, and biology). Over 30 research students helped to publish more than 20 articles. My students also gave research presentations at national and international conferences. Over 20 past research students are now employed by firms such as Google, Microsoft, Facebook, Amazon, Square, and Illumina. More than 20 students are pursuing Ph.D., M.D., or M.S. degrees at Columbia, Harvard, Yale, UCLA, UCSD, and the University of Chicago.
This summer, we will tackle several exciting projects using an extensive array of computational techniques, including machine learning, mathematical modeling, statistical analysis, and network science. Please keep in mind that I pair students with different backgrounds. Thus, knowing the biology underlying these problems is neither expected nor required for CS/Math students. Similarly, knowing how to code is neither expected nor required for biology students.
Project 1. Prediction of Human Addiction and Depression Disorders using Machine Learning
Mood disorders such as addiction and depression affect a sizable portion of the population today. Several studies have discovered a link between addiction and depression, and misalignments in human circadian rhythms caused by variations in circadian clock genes. Machine learning and statistical analysis will be used in this project to discover the most important genetic variations and clinical factors for addiction and depression. Then, using these risk factors, we will build a predictive machine learning classifier for addiction and depression.
Project 2. Analysis of the Gut Microbiome in College-Aged Student Athletes Who Have Recently Experienced Concussion: A Machine Learning and Network Analysis Approach
The gut microbiome is the community of billions of bacteria and other microorganisms of hundreds of different types found in the digestive tract. There is clear evidence that the bacterial composition of the gut microbiome affects brain activity and behavior. Furthermore, traumatic brain injuries (TBI) and other challenges to the brain can alter the composition of the microbes within the gut. In this study, we will use machine learning and network analyses to investigate the temporal relationship between concussion symptoms and gut bacterial composition to determine whether there is a link between microbiome changes and recovery from TBI.
Project 3. SEALNET: Facial Recognition Software for Ecological Studies of Harbor Seals
Seals are a top predator in marine coastal ecosystems and a key ecosystem regulator. Harbor seals occupy diverse climatic zones and environments and are widely considered the most successful pinniped species. Their high dispersal ability suggests limited population structure. Still, recent genetic and tagging studies suggest significant structure in populations—an indication that there is much to be learned from careful, in-depth studies of local populations. We recently developed a seal facial recognition software (SEALNET). In this project, we will improve the accuracy of our software, as well as automate and simplify the process to improve use of the software by others. Using our software, we will also find the structure of a population of harbor seals in Casco Bay, Maine.
Project 4. Design Principles of Adaptation Mechanisms in Biological Systems
Cells selectively respond to external stimuli to maintain cellular homeostasis by making use of different regulatory mechanisms. In this project, we will study the design principles of a group of distinct evolutionarily conserved biological mechanisms that allow cells to adapt to external stimuli. We will devise mathematical modeling and computer programming to compare and contrast responses of these adaptation mechanisms to a time-dependent external signal. Our analysis will exemplify the importance of computational modeling in the analysis of biological mechanisms.
Project 5. Effects of Dog Breed and Age on Thermoregulation of K9 Athletes
In animals selected for differing body masses, the amount of energy spent on thermoregulation may differ. Because of the large size disparity in dogs and little genetic variation, the domestic dog is the ideal model for assessing how body mass and metabolism have co-evolved. In this project, we will utilize machine learning and statistical analysis to understand the relationship between body temperature and heat dissipation in different types of canine elite athletes.
Student Qualifications
Computer science students must have taken COSC 102 and/or COSC 202. Math students must have taken MATH 260 and/or MATH 376. Biology students must have strong literature search and writing skills.
Number of Student Researchers
Four students
Project Length
10 weeks