Seminars

Molecular Biometry Cluster Candidate

Exact Bayesian Recursions in Bioinformatics

Charles (Chip) Lawrence, Ph.D., Bioinformatics Laboratory, Wadsworth Center & Computer Science Department, RPI

Thursday, March 13, 2003, 4 p.m.

1221 Computer Sciences and Statistics Center (CSSC), 1210 West Dayton Street

ABSTRACT

Heterogeneity in DNA composition impedes computational detection of transcription factor binding sites, and mRNA molecules often exist in an ensemble of structures. Dynamic programming algorithms are available to find optima for these problems but these optima yield point estimates that only partially address inferences of interest. For example RNA structural ensembles are characterized by a single solution by these algorithms. Fortunately because dynamic programming recursions can also be used to obtain large combinatorial sums, Bayesian inferences are accessible to address these shortcomings. Furthermore, when sums over the remaining variables can be completed simultaneously, direct Bayesian inferences on all variables can be obtained. I’ll discuss Bayesian inference procedures for the two biology problems mentioned above, their application to studies of transcription regulation, and the design of antisense and RNAi targets.

Back to General Departmental Seminar Series