Computational BiologyMany complex and varied phenomena in biology and medicine can be modeled with probabilistic or combinatorial algorithms. A confluence of tools from mathematics, computer science, and statistics, as well as an intimate knowledge of the application domain is required to build informative algorithmic models.
Our goal is to design algorithmic models to answer questions of import in biology and medicine.
The genomes of people, while largely identical due to evolution, contain variations. It is thought that the differences between people (hair color, eye color, height, disease, etc.) are due to contributions from the environment and genetics. Genomic variation is inherited and created through mutation during meiosis. This variation leaves a signature in people's genomes that can be useful for genealogical reconstruction, forensics, and even ecology.
Given the genomes of related people or animals, can we reconstruct the relationships that gave rise to those genomes? Do we need non-genomic information, such as birth/death records, to accurately reconstruct relationships?
For a particular genetic disease, medical doctors wish to scan a patient's genome for genomic variants that contribute to the risk of contracting that disease. In order to 'personalize' our understanding of genetic risk, we first perform studies on many people, in a population or family, to identify these variants.
Given the genomes of related people, how do we detect genomic sites that are associated with disease? How do we compute the statistics that tell us the relevance of each variant? The goal is to identify genetic variants that are present when the disease is present and absent with the disease is absent (or vise versa).
The genome is the vehicle of inheritance, but molecules such as RNA and proteins are the vehicles by which the genetic code is expressed into cellular activity and, through biological systems, contribute to producing the traits of the organism. Understanding the function of RNA will help us better understand the biological systems to which they contribute.
RNA molecules have a dynamic shape that can be modeled in two dimensions. An RNA's shape determines what other molecules it can interact with and changes in shape can encode information. For example, riboswitches bind metabolites which influence their shape and future interactions with other molecules.
Can we model the dynamic shape of RNA molecules? Can we determine which RNA sequences act as switches?
Data Privacy & Security
Due to legal and ethical concerns regarding data privacy (i.e., HIPAA laws and IRB considerations), computer security is a critical concern for health informatics and for computational biology research. Data must be handled in a privacy-preserving manner in order to respect patients and research subjects.