dVAR is a suite of software tools that 1) integrate high-dimensonal genomic data collected from large-scale public resources, including ENCODE, Roadmap Epigenomes, FANTOM5, and various conservation scores, 2) perform de novo pattern discovery from these integrated genomics data, and then 3) score the importance of noncoding variants across the whole genome taking advantages of the learned patterns.
The usage of dVAR is described in the README.
We also pre-computed the annotations of noncoding variants across the whole genome and the pre-computed annotations can be downloaded below. The format of the annotation is in the README file.
The source codes, along with the pre-computed annotation, can be DOWNLOADED here (Note that the file size is over 15GB).