Statistical principle-based approach for recognizing and normalizing microRNAs described in scientific literature

Area of Science:

• Bioinformatics
• Computational Biology
• Text Mining

Background:

• Scientific literature contains valuable microRNA (miRNA) information crucial for understanding biological pathways.
• Extracting miRNA mentions from both main text and figure captions is essential for comprehensive literature analysis.
• Accurate identification and normalization of miRNAs are challenging due to complex nomenclature.

Purpose of the Study:

• To develop a statistical principle-based method for recognizing and normalizing microRNA (miRNA) mentions in scientific literature.
• To link identified miRNAs to their corresponding identifiers in the Rfam database.
• To provide an interpretable knowledge representation for miRNA identification.

Main Methods:

• A statistical principle-based approach integrating pattern, dictionary, and supervised learning methods.
• Development of a miRNA recognition and normalization pipeline.
• Evaluation on the BioCreative VI Bio-ID corpus and an extended miRNA interaction corpus with Rfam normalization data.

Main Results:

• A robust method for identifying and normalizing miRNA mentions in scientific texts.
• Successful linking of identified miRNAs to the Rfam database.
• Demonstrated performance on established and extended benchmark datasets.

Conclusions:

• The proposed method offers an integrated solution for miRNA identification and normalization in scientific literature.
• Understanding miRNA nomenclature ambiguity is key to improving recognition accuracy.
• Highlights existing challenges and research gaps in automated miRNA literature mining.