← Back

Dr. Celia Greenwood

Event Start: Jan 23, 2019 1:00 pm
Location: Botterell Hall, Room B139

Dr. Celia Greenwood
Senior Investigator, Lady Davis Institute
Professor, McGill University
Montreal, QC, Canada

Two stories of genes, environment, and penalization

When exploring relationships between a trait (or a disease), one or more environmental factors, and a high dimensional genetic or genomic measure, we can make various assumptions about how to model and interpret the relationships.  I will tell two stories: one where a simple model goes a long way to explaining the trait of interest, and another where a complex model is needed. In the first story, a lasso-penalized model built on genome-wide genotyping data enables good prediction of fracture and bone mineral density, and is completely uncorrelated with known lifestyle predictors of bone mineral density.  In the second story, to model relationships between DNA methylation (measured via bisulfite sequencing) and phenotypes or covariates, we built a rich flexible model that allows for the effects, on the methylation levels, of multiple covariates to vary smoothly along small genomic regions. This method also allows for sequencing errors and can adjust for variability in cell type mixture. Here, penalization introduces smoothness for each covariate effect.  We have started exploring associations between rheumatoid arthritis, cell type, and methylation with this new model.
Wednesday, January 23rd, 2019
1:00 - 2:00 pm
Botterell Hall, Room B139