CAGE libraries were prepared from 5 μg of total RNA purified from 2×106 HeLa cells using the Purelink mini kit (Ambion) with 1% 2-Mercaptoethanol (Sigma) and on-column DNAse I treatment (Ambion) as recommended by manufacturer. CAGE libraries were prepared according as described previously49. Prior to sequencing four libraries with different barcodes were pooled and applied to the same sequencing lane. The libraries were sequenced on a HiSeq2000 instrument (Illumina). To compensate for the low complexity in 5’end of the CAGE libraries 30% Phi-X spike-in were added to each sequencing lane as recommended by Illumina. CAGE reads were assigned to their respective originating sample according to identically matching barcodes. Assigned reads were trimmed to remove linker sequences and subsequently filtered for a minimum sequencing quality of 30 in 50% of the bases using the FASTX-Toolkit (http://hannonlab.cshl.edu/fastx_toolkit/). Mapping to the human genome (hg19) was performed using Bowtie50 (version 0.12.7), allowing for multiple good alignments and subsequently filtering for uniquely mapping reads. Reads that mapped to unplaced chromosome patches or chrM were discarded.
