(A) Y-axis depicts the number of unique plasmid-derived mRNA fragments reported in a given STARR-seq (Arnold et al., 2013; Arnold et al., 2014; Zabidi et al., 2015), MPRA (Melnikov et al., 2012; Patwardhan et al., 2012; Tewhey et al., 2016), or CAP-STARR (Vanhille et al., 2015; Verfaillie et al., 2016) experiment. Each dot represents a replicate transfection (in cases where data on individual replicates were available). Published STARR-seq, MPRA, or CAP-STARR experiments that did not report the number of unique plasmid-derived mRNA fragments per replicate are not included. (B) Retransforming a given plasmid pool results in almost no loss in diversity. We retransformed a small amount (<300 ng) of the plasmid pool described in the main text into 300 ul of electrocompetent GT115 E. coli, and sequenced the inserts from the retransformed sample. For a given number of sampled fragments (x-axis), the retransformed sample and the post-transfection DNA-seq plasmid libraries from the main experiment have a comparable number of unique fragments (defined as uniquely mapped fragments with unique start and end positions). Each solid line represents an individual replicate, and the dashed line shows the line where x = y. The retransformed plasmid DNA library appears slightly more diverse than the methylated and unmethylated condition libraries, presumably because (i) the methylated and unmethylated condition libraries were sequenced post-transfection (which likely results in some loss of diversity) and (ii) the retransformed plasmid was supercoiled DNA rather than ligation product, likely resulting in a more efficient transfection. Importantly, almost all (98.17%) of fragments observed in the retransformed library are also present in methylated and unmethylated condition libraries.