FPfilter

FPfilter filters false-positive variant calls from whole-genome sequencing (WGS) data to improve the accuracy of variant calling pipelines such as BWA+GATK for clinical and research genomics.

Key Features:

• False-Positive Specificity: Prioritizes removal of false positives while minimizing loss of true positives compared to GATK hard filtering (GATK-HF).
• Model-Based Classification: Uses a classification model that distinguishes true positives and false positives by analyzing distinct patterns of heterozygous and homozygous mutations.
• Depth Gradient Analysis: Incorporates depth-gradient analysis using the Platinum Genome (PT) as a mutation reference and deep-sequenced datasets to assess the impact of sequencing depth on variant-calling accuracy.
• Performance Metrics: Demonstrates higher FP/TP filtration ratio and improved F-measure compared with GATK-HF on evaluated datasets.

Scientific Applications:

• Clinical genomics: Reduces false positives in diagnostic and treatment-planning workflows to improve reliability of variant interpretation.
• Research genomics: Enhances accuracy of WGS-based studies by refining variant call sets for downstream analyses.

Methodology:

FPfilter development and evaluation used the Platinum Genome (PT) as a mutation reference, analyzed depth gradients from NA12878 and deep-sequenced datasets, and assessed performance on NA12877 and NA24385 with comparisons to GATK-HF.