Dr. Dampier is a bioinformatician trained with an engineering mindset. He uses these principles to study the effects of HIV genetic on patient progression and the targeting of anti-HIV-1 genetic cure therapies. He utilizes 3^rd generation sequencing technologies like the Nanopore, a hand-held sequencing device, to investigate HIV genetic variation. He also studies how deep learning models can be used to answer biological questions that range from sequence to phenotype relationships to predicting future genetic variability.

Patent Approvals

• Composition and methods for HIV quasi-species excision from HIV-1-infected patients (U.S. Provisional patent application No. 62/084,182). Inventors: Brian Wigdahl, Michael R. Nonnemacher, and William Dampier. Filed November 25, 2014.
• Novel methods of determining clinical disease severity in HIV patients (U.S. Provisional patent application No. 62/193,861). Inventors: Brian Wigdahl, Michael R. Nonnemacher, William Dampier, and Vanessa Pirrone. Filed July 17, 2015.

Improvement of computational tools for the exploration and analysis of biological data. As part of my research I have contributed to numerous open source projects in an effort to improve their ability to handle data common to biological analysis. During my PhD and postdoctoral time, I began contributing to the Python scikit-learn project (scikit-learn.org/stable). This project aims to create a unified interface for a large collection of machine learning techniques. The unified interface allows a programmer to focus on the research question as opposed to the low-level implementation of these techniques.

As part of the team I have improved the handling of missing data throughout the toolset. This support is critical as many high-throughput biological experiments, such as microarrays, produce datasets with missing data points.

As part of my current research I have joined the Glue visualization project (glueviz.org). This tool was originally developed at Harvard’s Center for Astrophysics to visualize large, multi-dimensional astrophysics datasets. The tool natively produces multiple plot types; the novel aspect is that the plots are linked such that highlighting a particular region of interest propagates to all other graphs. I have been entirely responsible for integrating biological data-types and methods including, but not limited to: single and multiple sequence alignments, dates, categorical variables, viral deep sequencing analysis, and PCR primer design. I will continue to contribute to open source tools improve their usefulness for biological researchers.

First analysis of the CRISPR/Cas9 on mixed HIV-1 quasispecies. Using the CRISPR/Cas9 gene editing technology to excise HIV from latently infected cells is a promising strategy for curing HIV. However, the targeting strategy is non-trivial as the guide RNAs has non-linear binding potentials and HIV exists as a cloud of similar, yet distinct, genomes with low-frequency variants scattered throughout the genome. It is important to take both of these into consideration when designing guide RNAs to affect a sterilizing cure in which all quasispecies are excised. I was the first person to publish on the targeting ability of guide RNAs in a mixed population derived from deep sequencing of a patient’s HIV-1 quasispecies. This work has been extended to explore more patients as well as longitudinal effectiveness of guide RNAs within the same patient.

Exploration of pairwise correlated genome-wide changes in HIV. Traditionally, mutations in HIV have been explored on a gene-by-gene basis. However, the virus genome must act as a concerted whole to maximize its infectious potential and as such these correlations should be readily visible in the sequence. I have specifically examined how mutations in the V3 region of the gp120 envelope gene are correlated with mutations across the genome. Mutations in the V3 region can change the protein's affinity for either the CXCR4 and CCR5 co-receptors and as such can affect which cell-type the virus infects. My analyses have shown that there are mutations in the viral promoter (LTR) as well as the trans-activating factor (Tat) that are correlated with viral tropism.

Development of computational techniques for evaluating the effectiveness of HIV therapy. The capstone of my PhD research project was the implementation of an algorithm for using HIV sequence to estimate the effectiveness of anti-retroviral therapy (ART). Instead of using experimentally determined resistance sites, information not known for all therapies, I employed a machine-learning algorithm to learn from the presence or absence of eukaryotic linear motifs (ELMs). These are short (3-5 amino acid) segments that are known to correspond to functional elements of proteins such as phosphorylation sites, sumoylation motifs, and cell-tracking signals. My trained algorithm had similar F-scores to those using known resistance sites implemented by the Stanford HIV Database. This method was further expanded to predict human-HIV binding networks by looking at commonly shared motifs in human protein binding partners and extending those to estimating the potential to bind with HIV proteins sharing similar patterns of ELM motifs. This method was further employed on a non-HIV drug sensitivity problem when it was entered in the NCI DREAM competition for predicting the synergistic nature of drug pairs and ranked in the top 25%.

Determination of pathways common to multiple disease from publically available transcriptomic data. As part of my PhD and postdoctoral research I was involved in numerous projects that combined microarray data from multiple studies to explore the commonalities between similar diseases. Using a microarray dataset consisting of nearly 6,000 microarray samples from 84 laboratories composed of healthy and cancerous tissue from 13 tissue types, we found that many cancer types disrupt similar cellular pathways irrespective of source tissue. In collaboration with computational researchers at GlaxoSmithKline we used a similar technique to examine seven respiratory viruses. We found that signaling pathways such as the interferon-gamma and CD40 signaling were differentially regulated in 5 of the 7 viruses. My research has shown that combining publically available microarray dataset from multiple sources allows us to answer questions that were not originally envisioned by the researchers.

Exploring the DNA binding properties of DACH1 and CyclinD1 in mice. As part of my postdoctoral research, I was involved in numerous projects that explored the ability of DACH1 and CyclinD1 to control gene expression through DNA binding. I was directly involved in the computational analysis of chromatin immunoprecipitation sequencing (ChIP-Seq) data. Using traditional computation tools I was able to reveal that these transcription factors control a large fraction of genes related to breast cancer progression. Further gel-shifting assays confirmed many of these predicted results.