RNA-Seq compresses better than other sequencing platforms. Each benchmark dataset was randomly subset to the lowest common denominator number of reads (24 million) and read length (36 bases). Subsequently, Oculus computed the unique read fraction for each dataset using the reverse-complement option. For data with paired-ends available, 12 million pairs were used to computer %unique reads. RNA-Seq #1, Exome #1, and ChIP-Seq #1-2 did not have available paired-end data.