DGW: an exploratory data analysis tool for clustering and visualisation of epigenomic marks

Area of Science:

• Genomics
• Epigenomics
• Bioinformatics

Background:

• Functional genomics and epigenomics heavily rely on sequencing methods like ChIP-seq for DNA-protein interaction detection.
• These methods generate large, high-dimensional datasets with complex structures, such as multi-modal peaks over extended genomic regions.
• Existing visualization and data exploration tools have limited capabilities in handling these complex features.

Purpose of the Study:

• To introduce DGW, an open-source software package for simultaneous alignment and clustering of multiple epigenomic marks.
• To provide a tool that can effectively capture the intricate structure of epigenomic marks in large datasets.

Main Methods:

• The DGW package utilizes Dynamic Time Warping (DTW) for adaptive rescaling and alignment of genomic distances.
• This DTW approach enables the grouping of genomic regions with similar epigenomic mark patterns.

Main Results:

• DGW successfully performs simultaneous alignment and clustering of multiple epigenomic marks.
• The software demonstrates effectiveness in both simulation studies and analysis of real epigenomic data from the ENCODE project.

Conclusions:

• DGW automatically recognizes and aligns crucial genomic features, including transcription start sites and splicing sites, from histone mark data.
• The DGW software is available as an open-source Python package, promoting accessibility and further development.