An open-source multi-semantic annotation dataset and automated recognition tool for viral carcinogenesis factors

Area of Science:

• Oncology
• Virology
• Bioinformatics

Background:

• Understanding oncogenic viruses is crucial for cancer prevention and vaccine development.
• Viral carcinogenesis mechanisms involve genetic variations, lifestyle, and environmental factors.

Purpose of the Study:

• To build a comprehensive corpus of viral carcinogenic factors using NLP and expert knowledge.
• To analyze transcriptomic data for insights into the immune microenvironment and gene expression.
• To provide accessible tools and data for researchers and clinicians.

Main Methods:

• Developed a large-scale, semantically rich corpus of viral carcinogenic factors from 551,715 abstracts.
• Utilized natural language processing (NLP) and expert knowledge for entity recognition.
• Collected and analyzed transcriptomic data, including differential gene expression and immune microenvironment analysis.
• Created a semantic filter to enhance entity recognition.

Main Results:

• Compiled a dataset with 5,821,308 entities related to viral carcinogenesis.
• Identified key factors influencing viral carcinogenesis through data analysis.
• Established a visual knowledge platform and an open-source dataset for public access.

Conclusions:

• The study provides novel insights into viral carcinogenesis mechanisms.
• The developed resources facilitate rapid identification of research and clinical clues.
• This work supports the early prevention and control of virus-related cancers.