Literature based discovery of gene clusters using phylogenetic methods

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Biomedical literature contains valuable data for gene organization related to disease etiology and pathogenesis.
• Organizing genes associated with complex diseases like breast cancer is crucial for understanding disease mechanisms.

Purpose of the Study:

• To demonstrate the utility of existing phylogenetic methods for organizing breast cancer-associated genes.
• To compare the effectiveness of distance-based and character-based phylogenetic clustering methods.

Main Methods:

• Utilized MeSH annotations from over 35,000 MEDLINE articles to analyze 375 breast cancer-associated genes.
• Compared clustering performance using the Colless Imbalance Index (Ic) for distance-based methods and Maximum Parsimony (a character-based method).
• Evaluated the relevance of clustered genes, particularly those near BRCA1 and BRCA2, based on exclusive MeSH terms.

Main Results:

• Phylogenetic methods effectively organized breast cancer genes based on MeSH annotations.
• Both distance-based and character-based methods showed utility in clustering genes.
• Analysis of genes clustering around BRCA1 and BRCA2 revealed differences in relevance based on the clustering approach.

Conclusions:

• Existing phylogenetic methods and metrics are suitable for organizing genes according to knowledge in biomedical literature.
• This approach aids in structuring genetic information for disease research, particularly for complex diseases like breast cancer.
• The study highlights the potential of computational methods for knowledge discovery in large-scale biomedical text data.