Figure 4. Demonstration of the accuracy and precision of diPOLE.

(a) Euclidean distance estimates between probes at 20 bp (red circles) and 30 bp (blue circles) separation, respectively. Error bars are theoretical, assuming shot noise as only source of error. The expected Euclidean separation based on the dsDNA structure (black solid lines) is 6.8 nm and 10.2 nm, for 20 and 30 bp separation, respectively, with an uncertainty (dotted lines) because of minor and major groove interactions between fluorophores and dsDNA (Supplementary Fig. 12). We find weighted mean separations of, respectively, 8±1 and 12±1 nm (mean±theoretical s.e.m.; dashed lines with coloured area), which are in perfect agreement with the expected values (black solid lines) and in tantalizing agreement with the values expected if both probes are stuck in their extremal positions in the minor groove (dotted lines coinciding with dashed lines). (b) Same as a for relative angles. Error bars are experimentally calculated s.e.m. The expected relative angle between the probes (black solid line) is 74° in both samples because they differ in probe separation by a full repeat of the dsDNA. There is no discernable difference between the relative angles measured in the two samples. We found a mean orientation of 71±2° (mean±experimental s.e.m.; purple dashed line with purple area), which agrees with the expected value. The individual angle estimates scatter more than photon shot noise alone predicts. We therefore report here the experimental s.e.m. for realism. deg., degree.