Automated comparative auditing of NCIT genomic roles using NCBI

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Biomedical research has generated extensive human gene knowledge.
• The National Cancer Institute Thesaurus (NCIT) organizes this knowledge using concepts and roles.
• The rapid pace of genomic discovery necessitates regular auditing of terminological resources like the NCIT Gene hierarchy for errors.

Purpose of the Study:

• To present a comparative methodology for auditing the NCIT Gene hierarchy.
• To identify and suggest corrections for probable role errors within the NCIT Gene hierarchy.
• To assess the utility of automated comparative auditing for genomic knowledge repositories.

Main Methods:

• Developed web crawlers to access up-to-date data from both the NCIT and NCBI Entrez Gene databases.
• Implemented algorithms to compare gene-role information (chromosomal locations, biological processes) between the two databases.
• Utilized NCIT Biological Process hierarchy roles to identify discrepancies and probable gene-role errors.

Main Results:

• Discrepancies in gene chromosomal locations were striking and systematic, suggesting common origins.
• Identified numerous probable gene-role errors related to biological processes, exacerbated by multiple gene roles and process synonymy.
• The comparative auditing approach successfully uncovered a large number of potential errors and suggested corrections.

Conclusions:

• Automated comparative auditing is a promising technique for maintaining the accuracy of genomic knowledge repositories.
• The methodology effectively identifies probable errors in gene-role assignments within the NCIT.
• This approach facilitates the overall maintenance and reliability of the NCIT Gene hierarchy.