Extended Data Fig. 1: Experimental strategy used for the identification of mutated E/SEs in DLBCL.

H3K27Ac ChIP–seq data were generated in duplicate from 29 cell lines and 2 independent pools of human GC B cells, and used for the identification of active E/SEs based on the ROSE algorithm. The resulting list of E/SEs was intersected with the list of SNVs identified by WGS analysis in the same cell lines (matching each cell line to its own E/SEs) or in a panel of 93 de novo DLBCLs with matched normal DNA (Discovery Panel), to identify recurrently mutated E/SEs. An independent panel of 150 primary cases with WGS data and 169 primary cases with targeted-sanger sequencing data (Extension Panel) was then used to confirm the recurrent targeting of specific mutational hotspots identified in the BCL6-iSE, the BCL2-SE, and the CXCR4-SE.