Comparative Analysis for the Performance of Long-Read-Based Structural Variation Detection Pipelines in Tandem Repeat...

Area of Science:

• Genomics
• Bioinformatics
• Medical Genetics

Background:

• Structural variations (SVs) are increasingly linked to various diseases.
• Long-read sequencing technologies offer advantages for SV detection, particularly in complex genomic regions.
• Existing SV detection pipelines require comprehensive performance evaluation.

Purpose of the Study:

• To comprehensively evaluate and compare the performance of three common long-read SV detection pipelines: PBSV, Sniffles, and PBHoney.
• To specifically assess their efficacy in detecting SVs within tandem repeat regions (TRRs).
• To provide guidance for selecting appropriate long-read SV detection tools.

Main Methods:

• Utilized a robust benchmark dataset for germline SV detection as a gold standard.
• Evaluated precision, recall, and F1 scores for insertions and deletions.
• Compared pipeline performance both inside and outside of TRRs.

Main Results:

• All evaluated pipelines demonstrated superior performance in non-TRRs compared to TRRs.
• Sniffles achieved F1 scores of 0.60 within TRRs and 0.76 outside TRRs.
• PBSV performance was comparable to Sniffles and generally outperformed PBHoney.

Conclusions:

• Long-read SV detection pipelines perform better outside of tandem repeat regions.
• Findings assist researchers in choosing optimal pipelines for SV detection using long-read sequencing data.
• This study complements the application of long-read technologies in rare disease research.