StereoGene

StereoGene computes genome-wide correlations between continuous or interval genomic feature data to analyze spatial relationships among genomic annotations.

Key Features:

• Direct correlation of continuous data: Correlates continuous or interval genomic features without binarization to preserve quantitative signal.
• Kernel correlation technique: Uses kernel correlation to compute relationships among neighboring genomic positions and produces local correlation tracks.
• Partial correlation for confounder adjustment: Implements partial correlation to adjust for confounders such as variations in input DNA.
• Applicability to high-throughput datasets: Operates on datasets including ChIP-Seq, Human Epigenome Atlas, and FANTOM CAGE annotations mapped to a reference genome.
• Regulatory genomics insights: Detects changes in correlations between epigenomic features across developmental trajectories and tissue types.
• Discovery of novel spatial correlations: Identifies associations such as those between CAGE clusters and donor splice sites or poly(A) sites.
• Implementation: Implemented in C++.

Scientific Applications:

• Regulatory genomics: Quantifies spatial relationships among epigenomic features to investigate mechanisms of gene regulation.
• Comparative epigenomics: Analyzes changes in epigenomic feature correlations across developmental trajectories and multiple tissue types.
• ChIP-Seq and CAGE integration: Integrates ChIP-Seq and FANTOM CAGE datasets to map associations among genomic annotations.
• Transcript-associated signal discovery: Identifies spatial associations between CAGE clusters and donor splice sites or poly(A) sites.

Methodology:

Performs direct correlation of continuous or interval genomic data using kernel correlation to compute local correlation tracks and applies partial correlation to adjust for confounders such as input DNA variation.