Hidden biases in germline structural variant detection.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Genomic structural variations (SV) significantly influence genotypic and phenotypic traits.
• Accurate detection of SV from next-generation sequencing (NGS) data presents ongoing challenges.

Purpose of the Study:

• To investigate and identify the primary sources of analytical variability in structural variation (SV) detection using NGS data.
• To provide recommendations for reducing SV calling variability in large-scale genomic studies.

Main Methods:

• Sequencing DNA from a Chinese family quartet at three different centers in triplicate.
• Generating 288 derivative datasets using diverse analysis pipelines for comparison.
• Analyzing variability contributions from mapping methods, sequencing centers, and replicates.

Main Results:

• Mapping methods were the largest contributor to SV calling variability, followed by sequencing centers and replicates.
• SV supported by a single center or replicate showed high overlap with long-read SV call sets, indicating high false negative rates.
• SV calling variability persists even in genotyping, highlighting the influence of sequencing and preparation methods.

Conclusions:

• This study offers novel insights into the sources of variability in SV identification from NGS data.
• Challenges remain in SV calling for large cohorts, necessitating careful consideration of analytical pipelines and alignment methodologies.
• Recommendations are provided to mitigate SV calling variability and optimize alignment strategies.