BandHiC

BandHiC implements a memory-efficient banded storage strategy for organizing and analyzing Hi-C and Micro-C matrices at sub-kilobase resolution while maintaining NumPy-compatible operations.

Key Features:

• Banded Storage Strategy: Preserves only a configurable diagonal bandwidth of the dense contact matrix, reducing memory usage by up to 99% while maintaining fast random access and intuitive indexing operations.
• NumPy-Compatible Operations: Supports efficient vectorized operations optimized with NumPy for scalable analysis of ultra-high-resolution Hi-C datasets.
• Flexible Masking Mechanisms: Provides mechanisms to handle missing values, outliers, and unmappable regions.

Scientific Applications:

• High-Resolution Hi-C Analysis: Enables analysis of dense contact matrices from high-resolution Hi-C and Micro-C technologies that scale quadratically with resolution.
• Domain-Level Interaction Analysis: Facilitates investigation of chromatin domain-level interactions in 3D genome architecture.
• Ultra-High-Resolution Dataset Scalability: Supports scalable analysis of ultra-high-resolution Hi-C datasets with prohibitive memory demands using standard computational resources.

Methodology:

BandHiC adopts a banded storage strategy that preserves a configurable diagonal bandwidth of the dense contact matrix, with masking mechanisms for missing values, outliers, and unmappable regions, and vectorized operations optimized with NumPy.