Date of Award

Spring 5-15-2023

Author's School

Graduate School of Arts and Sciences

Author's Department

Biology & Biomedical Sciences (Molecular Genetics & Genomics)

Degree Name

Doctor of Philosophy (PhD)

Degree Type



In the era of advanced ability to perform complex genomic sequencing, precision oncology has been adopted as the ideal paradigm for optimization of outcomes for patients with cancer. However, despite technological advances in all aspects of the massively parallel sequencing pipeline, the application of precision oncology to every clinical workflow has been unattainable. Suboptimal adoption of custom medicine within oncology is attributable to the annotation bottleneck, which currently demands inordinate manual and computational requirements for completion. Alleviation of the annotation bottleneck requires co-development of bioinformatic strategies and analysis knowledgebanks to automate variant identification and variant annotation for clinical utility. The body of work presented here provides validated methods to alleviate the annotation bottleneck within the precision oncology pipeline. The introduction describes the specific aspects of the massively parallel sequencing pipeline that require development. Subsequently, we present three tools (DeepSVR, a Manual Review Standard Operating Procedure, and OpenCAP) that were developed to improve upon existing methods for variant identification and annotation. DeepSVR provides a machine learning approach to improve automated somatic variant calling by reducing false positives associated with sequencing pipelines that are observable by manual reviewers. The Manual Review Standard Operating Procedure provides a systemic and standardized approach for manual review of aligned sequencing reads for sequencing data with paired tumor and normal samples. Finally, the Open-sourced CIViC Annotation Pipeline (OpenCAP) serves as a software to create rationally designed clinical capture panels that are linked to clinical relevance summaries to improve library preparation and clinical annotation. The combined utility of these three tools for alleviation of the analysis bottleneck are demonstrated using a clinical example. Specifically, we developed a targeted clinical capture panel (MyeloSeq) to evaluate recurrent mutations observed in myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML). The MyeloSeq sequencing pipeline incorporated many of the tools described above for variant identification and annotation and provides a succinct output report for physician consumption. When surveying physicians who utilize the MyeloSeq panel, we observed that over 44% of physicians changed their treatment protocol based on the MyeloSeq results. This included 39 new therapeutics prescribes, 4 definitive diagnoses, and 13 changes in treatment plan (stem-cell transplant versus chemotherapy) based on prognostic indicators. This example demonstrates that the developed tools help alleviate the analysis bottleneck within precision oncology and will improve physician’s ability to integrate precision medicine into clinical workflow.


English (en)

Chair and Committee

Obi L. Griffith

Committee Members

Timothy J. Ley, Malachi Griffith, Lukas D. Wartman, Meagan A. Jacoby,