FUN-LDA scores for human genome assembly GRCh37

FUN-LDA is based on a Latent Dirichlet Allocation (LDA) model for predicting functional effects of non-coding genetic variants in a cell type and tissue-specific way by integrating diverse epigenetic annotations for specific cell types and tissues from large-scale genomics projects such as ENCODE and Roadmap Epigenomics. Using this unsupervised approach, we predict tissue-specific functional effects for every position in the human genome for 127 tissues and cell types in ENCODE and Roadmap Epigenomics. This online spreadsheet includes the information about the 127 Roadmap tissues in detail.

Format

The FUN-LDA scores are stored in the UCSC Genome Browser bigWig Track Format. 

To extract FUN-LDA scores, the bigWigAverageOverBed utility is required. It can be downloaded from the Genome Browser website at https://hgdownload.cse.ucsc.edu/admin/exe.

User should prepare a UCSC Genome Browser bed file, a tab-separated four-column file. The first column is the chromosome; the second is the zero-based coordinate of the position of interest; the third is that zero-based coordinate plus one; and the fourth is a unique identifier for the position. The command below is an example of extracting FUN-LDA scores in tissue E007 for the positions defined in a bed file, "input.bed".

bigWigAverageOverBed E007.valley9.c89.bigwig input.bed output.tab

It produces a six-column file, output.tab. The first column includes the unique position id defined in the input bed file. The last column, average over the covered bases, is the score for this position.

Reference

Daniel Backenroth, Zihuai He, Krzysztof Kiryluk, Valentina Boeva, Lynn Pethukova, Ekta Khurana, Angela Christiano, Joseph Buxbaum, Iuliana Ionita-Laza. FUN-LDA: A latent Dirichlet allocation model for predicting tissue-specific functional effects of noncoding variation: Methods and applications. American Journal of Human Genetics, 2018.