Sensitive and error-tolerant annotation of protein-coding DNA with BATH

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Accurate annotation of protein-coding DNA is crucial for understanding gene function.
• Existing tools may struggle with sequences containing errors like insertions and deletions (indels).
• Profile hidden Markov models (pHMMs) are powerful for sequence analysis but require specialized tools for annotation.

Purpose of the Study:

• To introduce BATH, a novel tool for highly sensitive annotation of protein-coding DNA.
• To improve the accuracy of DNA annotation in the presence of frameshift errors caused by indels.
• To simplify the workflow for pHMM-based translated sequence annotation.

Main Methods:

• BATH utilizes direct alignment of DNA sequences against protein databases or pHMMs.
• The tool is built upon the HMMER3 codebase.
• BATH incorporates novel frameshift-aware algorithms to detect nucleotide indels.

Main Results:

• BATH achieves accuracy comparable to HMMER3 for error-free sequences.
• BATH demonstrates superior accuracy compared to other tools for sequences with nucleotide indels.
• The tool provides a simplified input interface and interpretable output.

Conclusions:

• BATH is recommended for high-sensitivity DNA annotation, especially when frameshift errors are anticipated.
• The tool is particularly useful for analyzing long-read sequencing data and pseudogenes.
• BATH enhances the accuracy of protein-coding DNA annotation in the presence of sequencing errors.