VariantMetaCaller

VariantMetaCaller integrates outputs from multiple variant callers and aligners using Support Vector Machines to improve precision and sensitivity of variant detection in next-generation sequencing (NGS) data.

Key Features:

• Automated Fusion of Variant Calling Pipelines: Integrates measurement-related outputs from GATK UnifiedGenotyper, GATK HaplotypeCaller, FreeBayes, SAMtools and aligners such as BWA and Bowtie 2 to combine complementary information.
• Support Vector Machines (SVM) Integration: Uses Support Vector Machines to combine multiple data sources from variant calling processes and to estimate probabilities for each variant.
• Quantitative Precision-Based Filtering: Orders variants by computed probabilities and enables threshold adjustment to balance sensitivity and precision as an alternative to hard filtering or recalibration.
• Versatility Across Target Regions: Demonstrates applicability across target region sizes from a few hundred kilobases to whole exomes and to organisms lacking high-quality variant call sets.
• Application in Diverse Conditions: Supports quantitative filtering under broader conditions than traditional recalibration-based methods, useful when recalibration is impractical due to limited genomic datasets.

Scientific Applications:

• Improved variant detection in genomic studies: Enhances sensitivity and precision of variant calls in NGS analyses.
• Alternative to recalibration-dependent workflows: Provides a probabilistic filtering approach when high-quality reference datasets or recalibration are unavailable.
• Cross-species and targeted sequencing use: Applicable to human genomics, other organisms, and varying target region sizes including whole exomes.

Methodology:

Computationally integrates outputs from GATK UnifiedGenotyper, GATK HaplotypeCaller, FreeBayes, SAMtools and aligners BWA and Bowtie 2, applies Support Vector Machines to estimate per-variant probabilities, and performs quantitative probability-based filtering and ranking to balance sensitivity and precision.