LD Score regression

Individual active histone modifications were downloaded from the list of tissue types used previously (Finucane et al., 2015 (https://data.broadinstitute.org/alkesgroup/LDSCORE/) or - for mesenchymal isolated cell types which were not included in the Finucane analysis - the corresponding files from Roadmap itself (https://egg2.wustl.edu/roadmap/web_portal/). For the qq-plots, each annotation was run through stratified LD Score regression independently, including the baseline annotations and the Roadmap “control” annotation (Finucane et al., 2018), using the default parameters. The resulting enrichments were aggregated by cell type. For the “group” annotations showing tissue class enrichment, the set of cell type group annotations that was previously reported (Finucane et al., 2015) was augmented with the aggregate signal from mesenchymal lineage cells and tissues, and each was run through the stratified LD Score regression pipeline with the default parameters, using the baseline and Roadmap “control” annotations as above.