Composite Likelihood Methods - and their use in genetics. A composite likelihood function is an estimation function formed as a product of functions that individually are marginal or conditional likelihood functions. The use of maximum composite likelihood has gained traction as an approximate approach for addressing inference problems in cases where the full likelihood function cannot be evaluated efficiently using computation. In genetics in particular, composite likelihood has been used for mapping genes, estimating recombination rates, inferring natural selection, and inferring demographic parameters. This course will mix lectures, student led discussions of primary literature, and supervised research projects on composite likelihood methods. We will discuss recent literature form both the statistical and biological journals on theory relating to composite likelihood and its applications in genetics. This course is suitable for any graduate students with strong training in probability and statistics and an interest in learning more about both composite likelihood methods and genetics. No previous training in genetics will be assumed. May be taken for 1-3 units.
Course Format:
Prerequisites: Consent of instructor.
Credit option: Course may be repeated for credit.
Description: Special tutorial or seminar on selected topics.