SubPatCNV

SubPatCNV: Approximate Detection of Subgroup-Specific Copy-Number Variations

SubPatCNV implements an approximate subspace pattern mining algorithm to identify aberrant copy-number variation (CNV) regions specific to arbitrary sample subgroups in high-density array data, using spatial constraints and user-defined support thresholds.

Key Features:

• Approximate Subspace Pattern Mining: Detects consistent CNV subspace patterns within sample subsets using approximate association pattern mining.
• Spatial Constraints on CNV Probes: Applies positional constraints to CNV probe features to identify nearly identical deletions or insertions across subgroups.
• Support Threshold Filtering: Identifies aberrant CNV regions specific to subgroups exceeding a predefined minimum support.
• Scalability: Processes high-density array datasets and adapts to variable subgroup sizes.

Scientific Applications:

• Population-Specific Germline CNVs: Identified germline CNVs in HapMap samples across four distinct populations.
• Cancer Genomics: Detected large aberrant CNV events in TCGA ovarian cancer patient subgroups, highlighting regions enriched with cancer-related genes.

Methodology:

The method applies approximate association pattern mining under spatial constraints to high-density array-derived CNV probe data. It searches for nearly identical deletion or insertion patterns within subgroups that satisfy a minimum support threshold, enabling detection of subgroup-specific CNV regions across heterogeneous populations.