Figure 3.

Our ssNA modeling strategy consists of four steps. The first involves iterating the chain building procedure using a statistically-weighted dinucleotide suite library to populate a large structure pool. The pool is populated with 1000 models in each refinement round, while the final round utilizes 10 000 structures. In the next step, the SAXS profiles for all models in the pool are calculated. In step 3, the structure pool is fitted to input experimental SAXS data using ensemble optimization. This step selects sets of 20 structures from the pool whose calculated scattering profiles reconstitute the experimental data. This process is repeated 50 times in each refinement round, and 500 times in the final round. Finally, the selected structures are examined against the pool and a new set of suite weights is defined. This process is iterated until convergence.