Halvade Somatic

Halvade Somatic performs scalable somatic variant calling from whole-genome and whole-exome DNA sequencing data to enable detection of low-frequency tumor variants in cancer genomics.

Key Features:

• Scalability: Uses Apache Spark to parallelize data streams across multi-node and multi-core compute platforms for distributed processing.
• Comprehensive Pipeline: Implements GATK best practices including read alignment with BWA, read sorting, duplicate marking, and GATK base quality score recalibration followed by somatic variant calling.
• Variant Callers: Performs somatic calling with Mutect2 and supports Strelka2 as an alternative or complementary caller.
• Speed Optimization: Reduces runtimes substantially (e.g., from 84.5 h to 19.5 h on a single 36-core node, ≈4.3× speedup; to 1.36 h on 16 nodes, ≈14.4× additional speedup).
• Versatility: Supports both whole-genome sequencing and whole-exome sequencing data types.
• Deployment Flexibility: Can be executed on local clusters, Amazon EC2, Google Cloud, or via a Docker-based single-node deployment.

Scientific Applications:

• Cancer genomics: Detection and characterization of somatic single-nucleotide variants and small indels in tumor samples, including low-frequency variants.
• High-throughput sequencing analysis: Accelerated processing of large WGS/WES datasets to support timely tumor genetics studies and inform personalized treatment strategies.

Methodology:

Integrates BWA for alignment, read sorting, duplicate marking, GATK base quality score recalibration, and somatic calling with Mutect2 (and optionally Strelka2) within Apache Spark to optimize I/O and parallel processing following GATK best practices.