Parsing Immune Correlates of Protection Against SARS-CoV-2 from Biomedical Literature

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• The emergence of SARS-CoV-2 necessitates understanding immune correlates of protection (CoPs).
• The overwhelming volume of COVID-19 literature poses a challenge for researchers tracking CoP findings.
• Efficiently synthesizing information on CoPs, vaccines, and experimental models is crucial for scientific progress.

Purpose of the Study:

• To develop computational tools for identifying and extracting information on SARS-CoV-2 immune correlates of protection (CoPs).
• To create a user-friendly platform for visualizing and accessing up-to-date research on CoPs.
• To lay the groundwork for real-time informatics systems that deliver timely scientific insights.

Main Methods:

• A machine learning classifier was developed to identify research papers relevant to CoPs.
• A custom named entity recognition (NER) model was employed to extract key terms (CoPs, vaccines, assays, animal models).
• Extracted data were normalized and integrated into a visualization tool with publication and clinical trial links.

Main Results:

• Successfully developed and implemented machine learning and NER models for CoP literature analysis.
• Created a visualization tool that consolidates extracted information, offering researchers easy access to relevant data.
• Demonstrated the feasibility of using computational approaches to manage and present emerging research findings.

Conclusions:

• The developed tools provide a valuable resource for researchers studying SARS-CoV-2 immunity.
• This pilot project establishes a foundation for advanced, real-time informatics platforms in infectious disease research.
• Automated information extraction and visualization can accelerate scientific discovery in rapidly evolving fields like COVID-19 research.