BinSanity

BinSanity clusters metagenomic contigs into putative genomes using coverage and compositional signals to improve recovery of microbial genomes and link phylogeny to potential metabolisms.

Key Features:

• Affinity propagation clustering: Employs the affinity propagation (AP) algorithm for coverage-based clustering of assemblies.
• Coverage-based clustering: Clusters contigs based on coverage data to group sequences from the same source organisms.
• Compositional refinement: Refines clusters using tetranucleotide frequency and percent GC content.
• Separation of signals: Separates composition-based clustering from coverage-based clustering to reduce biases among closely related taxa.
• Bias mitigation: Addresses biases against low-coverage or low-abundance organisms and closely related strains that affect methods such as k-means, Gaussian mixture models, hierarchical clustering, neural networks, and two-way clustering.
• Performance metrics: Demonstrated higher precision, recall, and Adjusted Rand Index compared to five commonly implemented binning methods.
• Bin quality: Generates bins with high completion and low redundancy in environmental metagenomes.
• Unsupervised operation: Operates as an unsupervised clustering solution.

Scientific Applications:

• Environmental metagenomics: Recovery and binning of genomes from microbial communities in ecosystems with low biomass and limited cultured representatives.
• Metagenome-assembled genomes (MAGs): Generation of bins that align with established MAGs, showing high completion and low redundancy.
• Benchmarking: Evaluation on artificial metagenomes of varying sizes and complexities to compare precision, recall, and Adjusted Rand Index against other binning methods.

Methodology:

BinSanity applies affinity propagation (AP) clustering to coverage data, then performs compositional refinement using tetranucleotide frequency and percent GC content, separating composition- and coverage-based clustering and optimizing bins to contain multiple source organisms.