Livermore Metagenomics Analysis Toolkit

Livermore Metagenomics Analysis Toolkit assigns taxonomic labels to reads in large metagenomic datasets to produce comprehensive taxonomic profiles for biological samples.

Key Features:

• Scalable Performance: Shifts computational cost to an offline computation phase by creating a taxonomy/genome index to support large-scale classification.
• Kmer-Based Single-Pass Classification: Uses a kmer-based approach to analyze each read in a single pass for rapid taxonomic assignment.
• Read Binning for Downstream Analysis: Produces read bins that enable downstream metagenomic assembly and sensitive database searches on targeted subsets of reads.
• Efficient Large-Scale Classification: Demonstrated accurate taxonomic classification on very large datasets, exemplified by processing a 150 giga-base dataset.
• Implementation: Implemented in C++.

Scientific Applications:

• Diverse Microbial Profiling: Classifies viruses, prokaryotes, fungi, and protists within complex biological samples.
• Novel Organism Detection: Handles novel organisms to aid discovery of previously undetected microbial entities.
• Targeted Downstream Analyses: Enables selection of subsets of reads for more computationally intensive analyses such as assembly and sensitive database searches.

Methodology:

Creates a taxonomy/genome index in an offline computation phase; performs kmer-based single-pass analysis of each read and produces read bins; processed the 150 giga-base Tyrolean Iceman dataset in under 20 hours on a single node with 40 cores and large memory.