Mining locus tags in PubMed Central to improve microbial gene annotation

Area of Science:

• Microbial genomics
• Bioinformatics
• Scientific literature mining

Background:

• Millions of microbial gene identifiers in scientific literature are not incorporated into public databases.
• The Open Access (OA) subset of PubMed Central (PMC) offers a potential resource for gene annotations.
• Existing methods fail to systematically extract these valuable annotations.

Purpose of the Study:

• To develop an automated method for mining and extracting microbial gene locus tags from OA publications.
• To assess the utility of the OA subset of PMC as a source for microbial gene annotations.
• To create an R package, pmcXML, for efficient locus tag extraction.

Main Methods:

• Utilized the pmcXML R package to mine locus tags from full text, tables, and supplementary materials of OA publications.
• Extracted locus tags from 1835 OA publications across ten microbial genomes.
• Identified and expanded locus tag pairs defining genomic regions like operons and islands.

Main Results:

• Successfully extracted 30,891 locus tags from main text and 20,489 from tables across 1835 OA publications.
• Identified an additional 13,043 locus tags by expanding identified genomic region ranges.
• Analysis of *Burkholderia pseudomallei* revealed that locus tags in supplementary materials and non-OA publications constitute a significant portion of mentions.

Conclusions:

• The OA subset of PMC is a rich, underutilized resource for microbial gene locus tag annotation.
• The pmcXML R package effectively extracts locus tags, enhancing annotation databases, particularly for genomes with limited functional characterization.
• The developed R package provides access to valuable OA content not currently indexed by PMC searches.