ACCOST

ACCOST assigns statistical significance to differences in Hi-C contact counts to identify differential chromatin contacts while accounting for genomic distance effects.

Key Features:

• Statistical Modeling: Extends the DESeq statistical framework by adapting the 'size factors' concept to Hi-C contact counts to model sample-specific and distance-dependent effects.
• Genomic Distance Effect: Explicitly models and adjusts for the genomic distance effect that causes higher contact frequencies between proximal loci in Hi-C matrices.
• Unbiased Statistical Confidence: Provides unbiased estimates of statistical confidence as demonstrated on simulated and real Hi-C datasets.
• Comparative Performance: Demonstrates superior performance relative to diffHiC, FIND, and HiCcompare in comparative analyses.
• Implementation: Implemented in Python.

Scientific Applications:

• Cellular Differentiation: Detects changes in chromatin organization associated with cell-state transitions using differential Hi-C comparisons.
• Developmental Biology: Tracks evolution of genomic interactions during development through differential analysis of Hi-C data.
• Cancer Research: Identifies alterations in chromatin conformation that may be associated with oncogenesis or tumor progression.

Methodology:

Re-purposes the 'size factors' from DESeq to model the genomic distance effect and enable direct comparison of Hi-C contact counts across experiments; validation was performed on simulated and real Hi-C datasets.