CSSQ

CSSQ quantifies ChIP-seq signals to enable robust differential binding analysis while reducing noise and bias in epigenomic datasets.

Key Features:

• Statistical Modeling: Employs a finite mixture of Gaussians to model ChIP-seq data distributions.
• Noise and Bias Reduction: Applies an Anscombe transformation combined with k-means clustering and estimated maximum normalization to minimize noise and experimental bias.
• Non-Parametric Testing: Implements a non-parametric testing framework with comparisons under the null hypothesis via unaudited column permutation to support analysis with few replicates.
• Region-Level Analysis: Performs quantitative analysis on any defined regions within ChIP-seq datasets.

Scientific Applications:

• Differential Binding Analysis: Detects differential binding across ChIP-seq datasets with high confidence and low false discovery rate.
• Epigenomic Quantification: Quantifies epigenomic modifications to aid interpretation of epigenomic landscapes.

Methodology:

Finite mixture of Gaussians; Anscombe transformation; k-means clustering; estimated maximum normalization; non-parametric testing with unaudited column permutation for null comparisons; validated by simulations and benchmarking.