CompositeSearch: A Generalized Network Approach for Composite Gene Families Detection

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Gene evolution involves point mutations and the shuffling, fusion, and fission of genetic fragments.
• Sequence similarity can arise from common ancestry (homology) or shared genetic components.
• Analyzing large molecular datasets to disentangle these evolutionary processes is computationally intensive.

Purpose of the Study:

• To develop a memory-efficient, fast, and scalable method for detecting composite gene families in large datasets.
• To improve the accuracy, recall, and precision of composite and component gene family detection.
• To provide tools for critical biological analyses of gene family evolution.

Main Methods:

• CompositeSearch, a novel algorithm utilizing similarity networks.
• Generalization of similarity network approaches for enhanced gene family detection.
• Scalable processing of datasets typically containing millions of sequences.

Main Results:

• CompositeSearch demonstrates superior recall, accuracy, and precision compared to FusedTriplets and MosaicFinder.
• The method is memory-efficient, fast, and scalable for large-scale genomic data analysis.
• User-friendly quality descriptions of gene family distribution and conservation are provided.

Conclusions:

• CompositeSearch effectively disentangles homology from shared fragment evolution in large datasets.
• The tool enhances the biological analysis of gene family evolution and conservation.
• This method offers a significant advancement in analyzing complex genomic data.