normGAM

normGAM normalizes genome architecture mapping (GAM) data to correct systematic biases and improve chromatin interaction analyses, and it is implemented as an R package.

Key Features:

• Normalization Methods: Implements five normalization techniques: Normalized Linkage Disequilibrium (NLD), Vanilla Coverage (VC), Sequential Component Normalization (SCN), Iterative Correction and Eigenvector Decomposition (ICE), and Knight-Ruiz 2-norm (KR2).
• Bias Mitigation: Addresses window detection frequency, fragment length bias (newly identified and distinct from window detection frequency), mappability, and GC content in raw GAM data.
• Knight-Ruiz 2-norm (KR2): Includes a novel KR2 method developed for this package and evaluated for GAM normalization.
• Performance of VC and KR2: Vanilla Coverage (VC) and KR2 have demonstrated superior performance in correcting systematic biases.
• Validation with Orthogonal Data: Normalized GAM contact measures show increased alignment with normalized distances from fluorescence in situ hybridization (FISH) experiments.

Scientific Applications:

• GAM data normalization: Produces bias-corrected GAM contact matrices to improve downstream chromatin interaction analyses.
• Fragment length bias correction: Specifically targets a newly discovered fragment length bias analogous to biases observed in Hi-C data.
• Cross-platform harmonization: Enables comparison with Hi-C by maintaining higher correlation between KR2-normalized GAM and KR-normalized Hi-C on identical cell samples.
• Validation and benchmarking: Supports validation of GAM-derived spatial distances against FISH measurements.

Methodology:

Applies and evaluates five normalization methods (NLD, VC, SCN, ICE, KR2) to remove systematic biases, with comparative assessment of VC and KR2 performance against KR-normalized Hi-C correlations and FISH distance alignment.