Evaluation of alignment algorithms for discovery and identification of pathogens using RNA-Seq

Area of Science:

• Bioinformatics
• Virology
• Genomics

Background:

• Next-generation sequencing (NGS) offers powerful tools for viral discovery and characterization.
• Viruses exhibit high mutation rates, posing challenges for sequence alignment algorithms.
• Accurate viral sequence identification is crucial for understanding viral evolution and disease outbreaks.

Purpose of the Study:

• To assess the performance of eleven prominent alignment algorithms for characterizing mutated and non-mutated viral sequences in transcriptome data.
• To develop and utilize a realistic RNA-Seq simulation and evaluation framework (RiSER) for objective aligner assessment.
• To propose a combined score for ranking aligners based on precision, sensitivity, and alignment accuracy for viral characterization.

Main Methods:

• Development of the RiSER framework for simulating RNA-Seq data, including human and viral reads with varying mutation rates.
• Evaluation of eleven alignment algorithms: BLAST, BLAT, BWA, BWA-SW, BWA-MEM, BFAST, Bowtie2, Novoalign, GSNAP, SHRiMP2, and STAR.
• Application of a digital-subtraction-based viral identification framework and a new combined score for ranking aligner performance.

Main Results:

• Significant performance differences were observed among the eleven evaluated alignment algorithms.
• Sensitive aligners, including BLAST, BFAST, SHRiMP2, BWA-SW, and GSNAP, demonstrated the ability to accurately characterize divergent viral sequences with up to 15% mutation rate.
• The study suggests that different aligners may be optimal for distinct stages within a viral identification pipeline.

Conclusions:

• The choice of alignment algorithm critically impacts the accuracy of viral sequence characterization from NGS data.
• Highly sensitive aligners are recommended for identifying mutated and divergent viral sequences.
• The RiSER framework provides a valuable resource for objectively evaluating aligner performance in viral genomics research.