William Letsou earned his Ph.D. in Chemistry at the California Institute of Technology under Dr. Long Cai. His dissertation focused on the regulation of complex biological networks by noncommutative signaling molecules. Using techniques from enumerative combinatorics, he showed that temporal encoding involving a small number of regulators far surpasses the reachable space of traditional combinatorial/Boolean logic. Cells appear to use noncommutative logic when they sequester genes from the transcriptional machinery, for example, but it is a matter for future research how they deploy noncommutative signals in an autonomous fashion during development.
After Caltech, Letsou moved to St. Jude Children's Research Hospital in Memphis, Tenn., where he worked in the Department of Epidemiology and Cancer Control. There he developed computational tools to look for combinations of genetic variants associated with disease risk. Single-nucleotide polymorphisms (or SNPs) are locations in the genome at which alternate genetic variants confer small but significant risks for common diseases lacking an obvious causal mutation. SNPs form the basis of polygenic risk, a popular model which stratifies the population according to the number of risk variants an individual carries but does not take into account their combinatorial interactions. Letsou developed a pattern-mining algorithm to detect haplotypes—combinations of alleles on the same chromosome—enriched among individuals affected with disease. Using this method, he showed that a few top breast cancer risk SNPs are in fact poor tags for underlying rare, risk haplotypes of large effect, suggesting that the polygenic diseases may be more Mendelian than previously imagined.
Letsou's research at New York Tech aims to further the study of combinatorics in biology. With advances in parallel computing, it will be possible to study the association of hundreds of different variables with medically and biologically relevant outcomes. Equally important is developing sound statistical and physical models that make this information understandable to humans. Connecting the observed combinatorial associations with an underlying biological reality which may be noncommutative in nature is a long-term goal of his research program.
Recent Projects and Research
- Rare-haplotype associations with breast cancer risk, combinatorial pattern mining and algorithm development
- Theoretical biology of transcriptional regulation and autonomous development
Selected Publications
- Letsou, W., Wang, F., Moon, W., Im, C., Sapkota, Y., Robison, L. L. & Yasui, Y. Potential misrepresentation of inherited breast cancer risk by common germline alleles (2022). Submitted.
- Letsou, W. & Cai, L. Noncommutative biology: sequential regulation of complex networks. PLoS Comput Biol 12, e1005089 (2016).
Courses Taught at New York Tech
- BIOL 250: Biostatistics