BSCAMPP: Batch-Scaled Phylogenetic Placement on Large Trees.

Area of Science:

• Computational Biology
• Bioinformatics
• Phylogenetics

Background:

• Phylogenetic placement methods like EPA-ng and pplacer are crucial but struggle with large phylogenetic trees.
• The SCAMPP framework improved scalability by creating placement subtrees for each query sequence.
• SCAMPP's subtree generation is computationally intensive for numerous query sequences.

Purpose of the Study:

• To introduce Batch-SCAMPP (BSCAMPP), a novel technique to optimize phylogenetic placement for large datasets.
• To address the computational expense of SCAMPP when handling many query sequences.
• To improve the scalability and efficiency of phylogenetic placement.

Main Methods:

• Developed BSCAMPP to select a reduced set of placement subtrees based on query sequences.
• Integrated BSCAMPP with existing phylogenetic placement tools EPA-ng and pplacer.
• Evaluated BSCAMPP's performance on ultra-large backbone trees.

Main Results:

• BSCAMPP significantly reduces computation time compared to SCAMPP for both EPA-ng and pplacer.
• BSCAMPP demonstrates scalability to ultra-large backbone trees.
• BSCAMPP with EPA-ng is faster than with pplacer, while BSCAMPP with pplacer shows slightly higher accuracy but is slower.

Conclusions:

• BSCAMPP offers a computationally efficient and scalable solution for phylogenetic placement on massive datasets.
• The batching strategy effectively overcomes SCAMPP's limitations.
• BSCAMPP provides a valuable advancement for large-scale phylogenetic analyses.