Clustering analysis

Common peaks shared by two biological replicates in WT, ARID1A^-/-, and ARID1A^-/- ARID1B KD samples were merged and used to generate clusters based on peak shape in WT sample. k-means clustering was used to group according to peak tag density changes in ARID1A^-/- vs. WT cells. Tag densities from ARID1B KD and ARID1A^-/- ARID1B KD samples were then overlaid on these clusters. Heatmaps were created for accessible sites using annotatePeaks with -ghist parameter in HOMER. Window width was set as 1000 bp (±500 bp). Clustering by k-means was performed using Gene Cluster 3.0 (http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm) using Correlation (centered) for the Similarity Metric. Clusters were visualized using Java Treeview v1.1.6r4.