SPar-K

SPar-K partitions and analyzes archetypical chromatin signal profiles to identify chromatin architectures around functional genomic sites using a modified K-means clustering algorithm.

Key Features:

• Modified K-means algorithm: Uses a tailored version of K-means clustering adapted for genomic signal data.
• Signal-sensitive clustering: Handles vectors where sequence order, orientation and phase are significant for chromatin signal patterns.
• Handling data heterogeneity: Accommodates variation in signal profiles across genomic datasets to produce robust partitions.
• Misalignment correction: Allows limited misalignments of anchor points within genomic regions when clustering signals.
• Orientation flexibility: Detects phase shifts and orientation inversions by computing distances through shifting and flipping operations.

Scientific Applications:

• Chromatin architecture profiling: Partitioning genomic regions characterized by chromatin signal profiles to define recurring signal archetypes.
• Analysis around functional sites: Identifying archetypical structures around ChIP-seq peaks and other functional genomic anchors.
• Genome spatial-organization studies: Investigating how local chromatin signal patterns relate to genome organization and implications for gene regulation and expression.

Methodology:

Applies a modified K-means clustering algorithm on order-sensitive signal vectors, computing distances with shifting and flipping operations and allowing limited anchor-point misalignments.