
4 minute read
Viet Nguyen
Immune Subtypes as Biomarkers for Therapy Response in Glioblastoma Viet Nguyen
Mentor: Anna Joy Biology Department and Engineering
Introduction: Modifications to the scope of work Due to COVID 19 restrictions, the original proposal will not be performed since it requires in-person lab work. The student will now contribute to a bioinformatics project that can be performed remotely—immunotherapy for Glioblastoma (GB) patients. Despite hundreds of clinical trials over the last 40 years, little impact has been made on the median survival of GB patients. There is an urgent need for new approaches to treating these patients. Results from the GB phase 1 clinical trial of various new immunotherapy approaches, including checkpoint inhibitors delivered prior to surgery, oncolytic viruses, and tumor lysate pulsed dendritic cells have shown promise, but further trials are required to confirm activity. These studies indicate important obstacles remain since, in some cases, only a subset of patients responded, and for others, resistance limited a durable response. This highlights the need for a better understanding of GB-induced immune escape mechanisms, mechanisms of resistance to immunotherapies, and biomarkers of response. GB immune subtypes have distinct clinical outcomes: We used public single-cell RNA sequencing (scRNAseq) data from GB patient tumors to find genes expressed only in immune cells. We reasoned that these immune-specific genes can be used as a proxy for sequencing just the immune component of resected tumor tissue that contains tumor, blood vessel, immune, and normal brain cells. If true, then the immune-specific genes can be used on sequenced whole GBM tumors to discover information on its immune environment. We used clustering programs and immune-specific genes to find GB immune subtypes having distinct clinical outcomes. The data indicate that GB has distinct “types” of immune environments that have a significant impact on the clinical course of the tumor. Questions remain concerning: (1) the utility of immune subtypes as biomarkers of immunotherapy response (2) which GB genomic alterations contribute to immune suppression and therapy resistance, and (3) the immune composition of each subtype. We ultimately hope to develop and patent an algorithm that uses immune-specific genes as biomarkers for selecting effective immunotherapeutics. Our specific aims are: Specific Aim 1: Expand list of immune-specific genes. We will use additional public scRNAseq datasets of GB patient tumors to supplement the list of immune-specific genes. Specific Aim 2: Find GB genomic alterations associated with immune subtypes. We will use TCGA GB datasets containing mutations, epigenetic alterations, copy number variations, and gene expression of GB driver genes for > 200 GB tumors. We will find statistically significant enrichment of these genomic alterations in subtypes.
Methods Overview: On the first stage of this project, the main focus is to familiarize with the software that is used to in biomedical data processing. RStudio is used widely for three reasons. First, the chosen program was RStudio as it was deemed as an open-source software for scientific research with a reasonable learning curve for beginners. Furthermore, because it is a free tool and an open-source program, the software has a large number of community users that will constantly check and fix errors. Finally, RStudio is an appropriate tool to deal with a large set of data as it does not require the most advanced computing hardware and still provides detailed, high-quality graphics.
Materials and methods: The program was obtained for free online and installed into a personal laptop for use. To assist the process of learning how to operate RStudio, Swirl, an interactive teaching program in R was employed to guide the initial process. There were two options to choose when swirl is activated within 67
RStudio. The first option was to learn basic commands and functions in R programming, which includes 16 modules that instruct students to effectively navigate and work with the program. Each module took approximately 20-30 minutes to complete, which includes both basic introduction and practice problems to help acclimating learners to the software. The topics of these modules range from basic building blocks, sequence of numbers to logic and base graphics. After completing all of the introductory modules, the program would allow students to progress to the second option, which leads to the swirl course repository. This option provided a wide range of other programs that teach users how to apply the acquired knowledge in various scenarios. The courses that I completed were data analysis, getting and cleaning data, and statistical inference.
Results and Discussion: The decision to utilize swirl to guide new learners is appropriate as it was professionally structured and adjusted to novice users. Each module in the first part in swirl was built on the foundation of previously completed lessons and allow students to apply the newly acquired skills constantly. The integrated practice problems provide good repetition, which is crucial in learning a new skill. Finally, the lessons provided in the swirl repository were helpful in helping students to understand how to apply the skill in a research-related context.
Conclusion: The result, which is the learning outcome, in this case, was satisfactory as I was able to understand the basics of data sciences and can move on to the next steps of the research project. Even though some time is still needed to become completely proficient with RStudio, the training did provide all of the necessary information for me to progress.