Date of Award

Fall 2015

Document Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

Committee Director

Desh Ranjan

Committee Member

Mohammad Zubair

Committee Member

David T. Gauthier

Committee Member

Jing He


De-novo genome assembly from DNA fragments is primarily based on sequence overlap information. In addition, mate-pair reads or paired-end reads provide linking information for joining gaps and bridging repeat regions. Genome assemblers in general assemble long contiguous sequences (contigs) using both overlapping reads and linked reads until the assembly runs into an ambiguous repeat region. These contigs are further bridged into scaffolds using linked read information. However, errors can be made in both phases of assembly due to high error threshold of overlap acceptance and linking based on too few mate reads. Identical as well as similar repeat regions can often cause errors in overlap and mate-pair evidence. In addition, the problem of setting the correct threshold to minimize errors and optimize assembly of reads is not trivial and often requires a time-consuming trial and error process to obtain optimal results. The typical trial-and-error with multiple assembler, which can be computationally intensive, and is very inefficient, especially when users must learn how to use a wide variety of assemblers, many of which may be serial requiring long execution time and will not return usable or accurate results. Further, we show that the comparison of assembly results may not provide the users with a clear winner under all circumstances. Therefore, we propose a novel scaffolding tool, Correlative Algorithm for Repeat Placement (CARP), capable of joining short low error contigs using mate pair reads, computationally resolved repeat structures and synteny with one or more reference organisms. The CARP tool requires a set of repeat sequences such as insertion sequences (IS) that can be found computationally found without assembling the genome. Development of methods to identify such repeating regions directly from raw sequence reads or draft genomes led to the development of the ISQuest software package. ISQuest identifies bacterial ISs and their sequence elements—inverted and direct repeats—in raw read data or contigs using flexible search parameters. ISQuest is capable of finding ISs in hundreds of partially assembled genomes within hours; making it a valuable high-throughput tool for a global search of IS and repeat elements.

The CARP tool matches very low error contigs with strong overlap using the ambiguous partial repeat sequence at the ends of the contig annotated using the repeat sequences discovered using ISQuest. These matches are verified by synteny with genomes of one or more reference organisms. We show that the CARP tool can be used to verify low mate pair evidence regions, independently find new joins and significantly reduce the number of scaffolds. Finally, we are demonstrate a novel viewer that presents to the user the computationally derived joins along with the evidence used to make the joins. The viewer allows the user to independently assess their confidence in the joins made by the finishing tools and make an informed decision of whether to invest the resources necessary to confirm a particular portion of the assembly. Further, we allow users to manually record join evidence, re-order contigs, and track the assembly finishing process.