Ontological Vaccine Adjuvant Knowledge Presentation and Application in Adjuvant Design for Vaccine Development – UROP Summer Symposium 2021

Ontological Vaccine Adjuvant Knowledge Presentation and Application in Adjuvant Design for Vaccine Development

Amogh Madireddi

Amogh Madireddi

Pronouns: He/Him/His

UROP Fellowship: CCSFP, Washtenaw Technical Middle College
Research Mentor(s): Yongqun Oliver He, PhD
Research Mentor Institution/Department: Michigan Medicine, Departments of Lab Animal Medicine, Microbiology and Immunology, & Bioinformatics

Presentation Date: Wednesday, August 4th
Session: Session 3 (5pm-6:20pm EDT)
Breakout Room: Room 3
Presenter: 3

Event Link

Abstract

Many adjuvants (substances that increase immune response) are used in vaccine development. Some of their mechanisms have been studied extensively, others unknown. One challenge is how to collect and logically represent them so patterns can be identified and AI ready. Another challenge is how to use the patterns and information for machine learning. We aimed to first represent 25 vaccine adjuvants provided by our collaborators in the National Institutes of Health (NIH) using the Vaccine Ontology (VO), and then by analyzing these vaccine adjuvants together with over 100 adjuvants in the Vaxjo vaccine adjuvant database, we also aimed to design and recommend adjuvants optimal to stimulate protein-specific response against a particular pathogen (for a pathogen, you would know what immune response you need). To represent the NIH adjuvants, we store their info in a GitHub repository and also the VO, allowing future users to access these new adjuvants. Through work with NIH, we were able to represent additional vaccine adjuvants in the VO that can be used for future vaccine development. However, the designing and recommendation of adjuvants is more complex and took place in multiple steps. First, we utilized the adjuvants in the database and filtered out adjuvants that are not applicable to the user’s input using a binary tree and filtering algorithm. Next, we utilized the user’s protein sequence input to do a similarity search to look at adjuvant’s use for other pathogens. Through this action using machine learning classification, we are able to assign a score to each adjuvant that can be used to rank them in order of potential effectiveness for the user’s pathogen and vaccine target. This ultimately results in a ranked list of adjuvants for the user to experiment with for their given vaccine and pathogen. We also aim to develop a web application for the program so that it can be integrated into other programs on violinnet.org.

Authors: Amogh Madireddi, Yongqun Oliver He
Research Method: Computer Programming

lsa logoum logo