GATK hard filtering: tunable parameters to improve variant calling for next generation sequencing targeted gene panel...

Area of Science:

• Genomics and Bioinformatics
• Clinical Diagnostics
• Computational Biology

Background:

• Next-Generation Sequencing (NGS) offers advantages over Sanger sequencing for clinical applications.
• Standard variant calling software often uses fixed parameters unsuitable for diverse genes.
• GATK's parameter calibration typically requires extensive exome data, unavailable in diagnostic labs, necessitating manual hard filtering.

Purpose of the Study:

• To establish a procedure for optimizing GATK hard filtering parameters.
• To improve the accuracy of variant calling in a clinical diagnostic setting.
• To address the limitations of preset parameters in Next-Generation Sequencing variant analysis.

Main Methods:

• Simulated two datasets with varying coverages based on a real dataset's variant frequencies.
• Aligned simulated sequences using standard protocols.
• Employed regression trees to identify optimal GATK parameters and cutoff values for discriminating true and false variants.

Main Results:

• Successfully identified reliable parameters and cutoff values for variant call filtering.
• Analysis revealed that low-complexity flanking sequences correlate with a high rate of false positive calls.
• Demonstrated the effectiveness of simulation-based approaches for parameter optimization.

Conclusions:

• GATK hard filtering parameters can be effectively tailored using simulation studies.
• This approach enhances variant calling accuracy by focusing on the specific DNA region of interest.
• Provides a practical solution for diagnostic laboratories lacking large exome datasets for parameter calibration.