Novel read density distribution score shows possible aligner artefacts, when mapping a single chromosome

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Artificial data generation enhances accuracy and reduces computational time for evaluating aligners and peak callers.
• Mapping reads from a single chromosome is a strategy to decrease benchmarking time.

Purpose of the Study:

• To investigate potential artifacts introduced by single-chromosome mapping in aligner performance evaluation.
• To compare the accuracy of seven aligners using simulated data from single chromosomes versus whole genomes.

Main Methods:

• Comparison of seven aligners' performance on simulated benchmark data.
• Application of a novel read density distribution similarity measure to detect alignment artifacts.
• Analysis of mismatch statistics and mismatch frequency distributions.

Main Results:

• Common statistical methods are inadequate for comprehensive aligner performance evaluation.
• The novel read density similarity measure successfully revealed aligner performance artifacts.
• Mismatch statistics provided insights into alignment accuracy along reads.

Conclusions:

• Generating artificial data via single-chromosome mapping is a valid approach to reduce benchmarking time.
• The proposed quality assessment method identifies aligner shortcomings not detectable by conventional methods.
• These identified shortcomings can impact the alignment quality of real genomic data.