DNA flexibility is important both for fundamental biophysics and because DNA flexibility affects DNA packaging and regulation of gene expression through DNA looping. DNA flexibility has been studied with experiments ranging from biochemical ring closure or DNA looping experiments to AFM, crystallography, and tethered particle microscopy. Even so, the flexibility of DNA in vitro and in vivo remains controversial.

In an attempt to resolve this controversy, we have developed a high- throughput, internally controlled, comparative ligation methodology using a library constructed of 1023 distinct DNA sequences ranging in length from 119 to 219 base pairs via ligation of pools of synthetic DNA of different lengths and PCR. The design incorporated barcoding for redundant identification of each molecule, allowing for a ligation reaction to be performed on the entire library in one reaction mixture. Two DNA concentrations were used in separate reactions to promote either unimolecular cyclization or bimolecular ligation and thereby explore a wide range of cyclization efficiencies (J factors). Half of each reaction mixture was treated with BAL-31 to destroy non-cyclized molecules. All products were linearized by restriction digestion and Illumina indices were added. The initial library and reaction mixtures were sequenced in a single Illumina MiSeq run.

From roughly 15 million assembled reads, over 13 million were identified using software written to identify and sort our sequence library. Each molecule was counted for each condition. From our analysis we see no evidence of extreme bendability at short DNA lengths. At higher DNA concentrations where bimolecular products are produced more rapidly, we see oscillatory behavior as a function of length. In contrast, at lower concentrations where unimolecular products dominate, we observe no helical variation due to the ability for all molecules to cyclize given enough time. In order to determine J factors through cyclization, bimolecular products must also be counted. Given the constraints of this experiment, not all bimolecular products could be observed. Future experimentation can be performed to determine J factors across this size range, the results of which will improve coarse

grain modeling of DNA. Extension of this methodology should be applicable to DNA loops anchored by proteins.