David Spencer

Joined Program: 2004 (from UW Medical Scientist Training Program)
Previous Degree: B.S. Biochemistry, University of Washington
Olson Lab
dhs [ a t ] u.washington.edu

Research:

My research is focused on the development and application of a new strategy for identifying disease-causing mutations in the human genome. The basic premise of the method is that identification of disease-causing mutations will be more efficient if haplotypes from affected patients are compared to closely matched control sequences rather than the human reference genome, since the reduced level of background neutral variation that will result from these comparisons will correspond to fewer potential causative mutations that must subjected to further analysis.
I have examined the practicality of this method by calculating the expected divergence between a single “query” sequence and its closest match among a panel of control sequences for a simple model of sequence evolution. These calculations predict that the number of neutral polymorphisms can be reduced by a factor of 50 when a query haplotype is compared to its best match in a panel of 100 reference sequences, instead of a randomly chosen one. I have also examined the similarity between closely matched sequences in more realistic conditions, using coalescent simulations that include recombination and models of human demographic history, as well as a set of real sequences sampled from diverse human populations; both of these analyses suggest that excellent matching in real-world situations will be possible.

As an initial test of this method, I am searching for mutations in a 225-kilobase pair region containing the MAPT gene on chromosome 17q21 in two individuals with hereditary dementia. Linkage studies in the families of these individuals have implicated this locus, but no mutations in the coding regions of MAPT have been found despite the association of this gene with many other familial dementias with similar phenotypes. Haplotype matching in an extended region containing this entire gene will be performed, which will involve sequencing of the affected haplotypes from probands using targeted, haplospecific fosmid-based resequencing and selection of closely related control sequences. The discrepancies between the affected haplotypes and the closely matched controls will be considered potential disease-causing mutations.

Since haplotype matching reduces the level of background neutral variation, it enables unbiased mutation searches to be conducted on a larger scale without being overwhelmed by large numbers of potential disease-causing mutations. To this end, I plan to initiate a mutation search in a large, multi-megabase pair linkage interval using haplotype matching. Sequencing of a region of this size would be difficult using current technology, and so I will explore using next-generation sequencing platforms to sequence the affected haplotype.