ResLit

ResLit mines the antimicrobial resistance literature for resistance genes, mutations, organisms, and mechanisms, providing a comprehensive, evidence-linked repository of AMR findings at scale.

Key Features:

• Automated Literature Screening: BioMistral-7B screened 2 million candidate PubMed abstracts to identify 356,000 relevant papers.
• Multi-tier Full-text Retrieval: Multi-tier retrieval yielded 117,000 full texts for downstream extraction.
• Two-step Information Extraction: Qwen3-30B performed two-step extraction of resistance genes, mutations, organisms, and mechanisms from full texts.
• Cross-linked Evidence Tiers: ResLit contains 3,120 genes and 13,593 mutations, cross-linked to CARD, ResFinder, and NCBI Reference Gene Catalog across four evidence tiers.
• Community-driven Curation: The database supports community-driven curation of automated outputs and reference databases.

Scientific Applications:

• Antimicrobial Resistance Gene Discovery: Researchers can access 3,120 resistance genes identified through large-scale automated literature mining.
• Resistance Mutation Analysis: Users can investigate 13,593 mutations linked to antimicrobial resistance with evidence-based cross-references.
• Evidence-linked AMR Research: Scientists can utilize four evidence tiers cross-linked to CARD, ResFinder, and NCBI Reference Gene Catalog for validated findings.

Methodology:

ResLit employs BioMistral-7B for abstract screening of PubMed records, multi-tier retrieval for full-text acquisition, and Qwen3-30B for two-step extraction of resistance genes, mutations, organisms, and mechanisms.