Transcription start sites/clusters/regions (TSS/TSC/TSR) are given in GFF3 format (http://www.sequenceontology.org/gff3.shtml). File contains comments (lines starting with single hash "#"), meta-data (lines starting with double hash "##"), and predicted sites. Feature record contains 9 columns separated by tabs. Column 1: "seqid" The ID of the landmark used to establish the coordinate system for the current feature. In our case it is chromosome ID. Column 2: "source" This field describes method used for predicting the feature. In our case this is "CLUSTER". Column3: "type" Type of the feature. In our case these are TSS/TSC/TSR for transcription start sites/clusters/regions respectively. Column4: "start" Start position of the feature in 1-based integer coordinate system. Column5: "end" End position of the feature in 1-based integer coordinate system. Column6: "score" Not defined. Column7: "strand" Strand of the feature. "+" for positive strand and "-" for negative strand Column8: "phase" Not defined. Column9: "attributes" A list of feature attributes given in format tag=value. Tags: "ID" - unique ID of the feature. "Name" - name of the feature. "Parent_TSR" - name of parent TSR. "Parent_TSC" - name of parent TSC. "Representative_position" - representative position of TSC. "CpG_class" - CpG class of TSR