Fig 1: DeCET uncovers epigenetic patterns specific to myometrium and uterine leiomyoma subtypes(A) Schematic illustration of how the data tensor is decomposed into characteristic modes in each index space (condition (c), patient (pt), assay (a), and genomic location (loc). The bottom portion shows how the decomposition represents each ChIP-seq profile as a projection onto independent spatial patterns of histone modifications.(B) The projections of ChIP-seq datasets onto the first 10 HOSVD location vectors.(C) Unsupervised hierarchical clustering of the 21 patient-matched samples using 8 of the first 10 HOSVD projections for each assay (Figure 1B; STAR Methods). The columns correspond to an assay and location vector index pair. Leiomyoma tissues are labeled by observed mutations (Table S1): MED12 exon 2 mutations (MED12-mut), HMGA2 overexpression (HMGA2 high), biallelic loss of FH (FH low), or unknown if none of the above three were observed.See also Figures S1–S4 and Tables S1, S2, S3, S4, and S5.