Algorithms for Analyzing Human Genome Rearrangements

of “Algorithms for Analyzing Human Genome Rearrangements,” by Crystal L.Kahn, Ph.D., Brown University, May 2011. The human genome exhibits a rich structure resulting from a long history of genomicchanges, including single base-pair mutations and larger scale rearrangements such as in-versions, deletions, translocations, and duplications. The number and order of the genomicchanges that resulted in the present-day human genome is not known, but can sometimesbe inferred by comparison to the genomes of other species. In particular, genome rear-rangements are modeled as operations on signed strings of characters representing blocksof conserved sequences. Genome rearrangement distance measures quantify the similaritybetween two or more genome sequences by counting the minimum, or most likely, numberof rearrangement operations needed to transform one sequence into another. The devel-opment of efficient algorithms for computing genome rearrangement distances has beeninstrumental both in computing phylogenies for sets of known genetic sequences (such asgene families or the whole genomes of present-day species) and in constructing ancestralgenome sequences. In this thesis, we develop algorithms to study recent genome rearrangements in human andcancer genomes. We introduce a novel measure, called duplication distance, to quantifythe similarity between two genomic regions containing segmental duplications. We givean efficient algorithm to compute the duplication distance between a pair of signed stringsand provide several generalizations of duplication distance that also measure inversions anddeletions. We demonstrate the utility of the duplication distance measure in constructingthe evolutionary history of segmental duplications in the human genome using both parsi-mony and likelihood techniques. Further, motivated by recent cancer genome sequencingstudies, we present a new algorithm for the block ordering problem of inferring a wholegenome sequence from a partial assembly by maximizing its similarity to another genome. Algorithms for Analyzing Human Genome Rearrangements byCrystal L. KahnB. A., Amherst College, 2004Sc. M., Brown University, 2008 Submitted in partial fulfillment of the requirementsfor the degree of Doctor of Philosophyin the Program in Computer Science at Brown University. Providence, Rhode IslandMay 2011 c© Copyright 2008, 2009, 2010 by Crystal L. Kahn This dissertation by Crystal L. Kahn is accepted in its present form bythe Department of Computer Science as satisfying the dissertation requirementfor the degree of Doctor of Philosophy. DateBenjamin J. Raphael, Director Recommended to the Graduate Council DateSorin Istrail, Reader DateBernard Moret, Reader(École Polytechnique Fédérale de Lausanne) DateCharles Lawrence, ReaderApplied Mathematics Approved by the Graduate Council DatePeter WeberDean of the Graduate School