Home
Projects
..UTR Evolution
..Binding site variants
..Neutral expression evol.
..Protein mis-interactions
People
Publications
CisRegulation Info
Diversions
In collaboration with Henry Lu, Richard Lusk, and Wen-Hsiung Li.
We are developing methods for classifying protein interactions by type (indirect vs. direct) and strength (stable vs. transient). Currently, our predictions use a Bayesian framework to account for false positives and negatives. The model accounts for differences in protein-protein interaction assays: co-immunoprecipitation experiments detect stable complexes, while yeast two-hybrid experiments detect both stable and transient direct interactions, but require ectopic expression. We are using this framework to predict with good accuracy the probability of a true interaction, the type of interaction (direct vs. indirect) and strength of interaction (stable vs. transient).
We are characterizing protein pairs that are expressed at different times and locations in order to avoid interference between pathways. Because some protein interaction assays use ectopic expression and others use endogenous expression, we can identify protein interactions that are usually suppressed, but are strong when proteins are artificially coexpressed. We will experimentally test effect of these interactions that are usually suppressed on fitness by co-expression with plasmids in yeast.
We will identify gene duplicates that have different expression profiles, which has enabled paralogous pathways to diverge by reducing crosstalk. We will identify duplicate proteins whose interactions with proteins in paralogous pathways, normally suppressed, is strong when coexpressed, suggesting the duplicates have changed timing and location of protein expression to avoid interference. These proteins will have a negative fitness effect if they are artificially co-expressed and co-localized.
