APPLES

APPLES performs phylogenetic placement of DNA and protein sequences into pre-existing reference trees using kmer-based distances and least-squares optimization to enable scalable placement on large reference trees.

Key Features:

• Scalability: Handles reference trees with a very large number of leaves, tested up to 200,000.
• Speed and Efficiency: Achieves runtimes and memory usage an order of magnitude lower than traditional maximum likelihood (ML) methods on large datasets.
• Non-dependency on sequence alignment: Does not require aligned query sequences or assembled reference sequences, enabling placement from unassembled reads.
• kmer-based distance calculation: Uses kmer-based distances to compute pairwise sequence distances for placement without alignment.
• Accuracy with dense reference trees: Demonstrates improved placement accuracy when using dense reference trees compared to ML methods on sparser trees.

Scientific Applications:

• Phylogenetic updates: Scalable integration of new species data into existing phylogenies for taxonomy and tree updates.
• Sample identification: Identification of unknown query samples using DNA and protein data from (meta-)barcoding and other sequencing sources.
• Genome skimming applications: Placement of unassembled reads from genome skimming projects for sample identification.
• Metagenomic analyses: Placement of sequences from metagenomic datasets for community profiling and sample characterization.

Methodology:

APPLES performs distance-based phylogenetic placement using kmer-based distance calculations and least-squares optimization without requiring sequence alignment.