Literature mining and database annotation of protein phosphorylation using a rule-based system.

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Vast amounts of protein phosphorylation data exist in scientific literature, but are difficult to access and curate for databases.
• Manual literature curation is time-consuming and limits the comprehensive collection of this valuable data.

Purpose of the Study:

• To develop and evaluate a computational system for automated literature mining of protein phosphorylation.
• To facilitate the efficient extraction and annotation of protein phosphorylation information for biological databases.

Main Methods:

• Utilized a rule-based system named RLIMS-P (Rule-based LIterature Mining System for Protein Phosphorylation).
• Applied RLIMS-P to MEDLINE abstracts to identify and extract phosphorylation-related entities (kinases, substrates, sites).
• Evaluated system performance using an established, annotation-tagged literature corpus.

Main Results:

• RLIMS-P achieved high precision (91.4%) and recall (96.4%) for retrieving relevant phosphorylation papers.
• The system demonstrated high precision (97.9%) and recall (88.0%) for extracting protein substrates and phosphorylation sites.
• Combined high recall in paper retrieval with high precision in information extraction.

Conclusions:

• RLIMS-P effectively automates the extraction of protein phosphorylation data from scientific literature.
• The system significantly enhances the process of literature mining and database annotation for protein phosphorylation.
• This approach promises to accelerate the curation of crucial biological information.