Sequedex

Sequedex classifies short reads from shotgun metagenomics by exact matching to a precomputed set of peptide 10-mers to provide phylogenetic and functional assignments.

Key Features:

• High-Speed Processing: Processes approximately 6.6 gigabase pairs per hour on a single CPU compared to about 25 kilobase pairs per hour for BLASTX against the NR database.
• Exact Matching Methodology: Uses exact matches between reads and peptide 10-mers, with about 20 million 10-mer peptides identified across 403 representative bacterial genomes.
• Phylogenetic Classification: Assigns each peptide 10-mer as a signature to nodes in a phylogenetic reference tree (primarily based on RNA polymerase genes) and classifies fragments by the most specific node consistent with observed signature peptides.
• Sensitivity and Specificity: Exhibits sensitivity and specificity comparable to MEGAN when using BLASTX against NR, validated on synthetic data from newly sequenced soil-bacterium genomes and ten real soil metagenomics datasets.
• Signal-to-Noise Ratio: Achieves a signal-to-noise ratio of approximately 400 in real metagenomics data, enabling discrimination among environmental samples.
• Versatility with Read Lengths: Produces results comparable to existing techniques for reads longer than ~300 base pairs while maintaining robust performance with shorter reads.

Scientific Applications:

• Metagenomic classification in high-diversity environments: Classifies short reads from environments with high sequence diversity and few close reference sequences, such as soils.
• Large-scale environmental genomics: Enables rapid assignment of reads for inclusion in high-throughput environmental and ecological genomic analysis pipelines.
• Phylogenetic and functional genomics studies: Provides peptide-based signature assignments that support phylogenetic placement and functional inference in microbial community analyses.

Methodology:

Precomputes peptide 10-mers from representative bacterial genomes (~403 genomes, ~20 million peptides), assigns each 10-mer as a signature to nodes of a phylogenetic reference tree primarily based on RNA polymerase genes, and classifies reads by exact matching translated fragments to those peptide 10-mers to identify the most specific consistent phylogenetic node.