EXTRACT

EXTRACT extracts environment-related named entities from textual sample descriptions to annotate metagenomic records and other biological samples with standardized environmental metadata.

Key Features:

• Named Entity Recognition: Employs named entity recognition and text-mining to identify environments, organisms, tissues, and diseases in textual sample descriptions.
• Standard-compliant annotation: Aligns identified terms with established metadata standards to support standardized environmental metadata for metagenomic records.
• Evaluation (BioCreative V Interactive Annotation Task): Reported improvements in curation accuracy and a 15–25% increase in annotation speed compared with manual methods.
• Enhanced detection: Uses text-mining to help identify terms that might be overlooked during manual curation, increasing completeness of metadata.

Scientific Applications:

• Metagenomics sample annotation: Annotation of metagenomic records with standardized environmental metadata for downstream analysis.
• Microbial ecology: Support for microbial ecology studies through standardized environmental descriptors enabling comparison across samples and studies.
• Metadata curation and integration: Facilitation of large-scale metadata curation to improve data integration and analysis across datasets.

Methodology:

Combines published named entity recognition methods and text-mining techniques to extract relevant environment-related terms from text and align them with metadata standards.