Gene identification and protein classification in microbial metagenomic sequence data via incremental clustering

Area of Science:

• Computational biology
• Bioinformatics
• Genomics

Background:

• Metagenomic datasets present challenges for gene identification due to organismal variations (e.g., GC content, codon bias).
• Efficient protein family classification is crucial for analyzing large-scale sequence data.

Purpose of the Study:

• To present a computational improvement for identifying and classifying protein-coding genes in microbial metagenomic datasets.
• To enhance the speed and efficiency of protein clustering for large datasets.

Main Methods:

• An incremental clustering method was developed, improving upon a previous sequence clustering approach.
• The new method avoids computationally expensive all-against-all comparisons while retaining remote homology detection.

Main Results:

• The improved clustering approach effectively identifies and classifies protein-coding genes in prokaryotes, viruses, and intron-less eukaryotes.
• Evaluations demonstrate the method's efficiency in updating protein clusters with new genomic and metagenomic sequences.

Conclusions:

• The incremental clustering method significantly accelerates gene identification and protein family classification in metagenomic data.
• This approach facilitates the discovery of novel protein families and provides a foundation for further protein family studies.