Computational Methods for Designing Optimal Genomics-guided Viral Diagnostics

点击翻译按钮获取中文摘要

Project Summary/Abstract Viral genome sequencing is growing exponentially and cutting-edge molecular technologies, guided by genomic data, show great promise in detecting and responding to viruses. Yet we lack a computational framework that efficiently leverages viral data to design the nucleic or amino acid sequences applied by these technologies. The proposal provides a career development plan to (i) build computational techniques — algorithms, models, and software — that yield highly accurate diagnostic assays, with potential to outperform existing ones, and (ii) use the techniques to proactively design assays for detecting 1,000s of viruses. The project will first develop methods for designing optimal viral genome-informed diagnostics. The study will formulate objective functions that evaluate an assay’s performance across a distribution of anticipated viral targets. Combinatorial optimization algorithms and generative models, constructed in the study, will optimize the functions. The project will also develop datasets for training predictive models of assay performance, which are used in the objective functions, focusing on CRISPR-, amplification-, and antigen-based diagnostics. Preliminary experimental results suggest such models can render assays with exquisite sensitivity and specificity. The study will compare the algorithmically-designed assays to state-of-the-art tests for four viruses. With these methods, the project will design diagnostic assays that are species-specific and broadly effective across genomic diversity for all viruses known to infect vertebrates. The study will build a system to monitor the assays’ effectiveness against emerging viral genomic diversity and to continually update them as needed. To enable the broad adoption of these methods, the project will implement them efficiently in accessible software. The proposal aligns with a NIAID goal of improving diagnostics via data science. The methods developed here may also aid therapy and vaccine design, and will leave the world better prepared to combat viral outbreaks. The career development award will provide training for the candidate in applied areas of long-term interest to his career. The candidate has previous experience in developing computational methods and analyzing viral genomes. Through the award, he will gain new knowledge and skills in diagnostic applications, alongside formal and informal training in immunology, bioengineering, and related laboratory techniques. This training will help the candidate progress toward therapy and vaccine applications that could benefit from advanced computational methods. The Broad Institute provides a supportive environment for the candidate’s development, including career development workshops, research seminars aligned with the proposed plan, and opportunities to initiate collaborations with scientists having expertise complementary to the candidate’s. The research and training will help him form an independent research group focused on developing and applying computational methods to enable more effective microbial surveillance and response.