A modular strategy for generating starting conformations and data structures of polynucleotide helices for potential energy calculations

We describe a simple and rapid algorithm for generating data structures and starting coordinates of polynucleotides for potential energy calculations. The algorithm is tailored to investigations in cartesian coordinate, rather than dihedral angle, space. First, instead of a tree structure for molecular design, we set up a helix from a simple list of bonds for the basic DNA subunits (sugar, phosphate, and bases). Second, instead of using successive transformations to obtain a set of coordinates in one reference frame, we apply a simple “matching” routine to patch DNA subunits. Third, we avoid ring closure and geometry optimization by allowing deviations from equilibrium values only for PO3′ bond lengths and O5′PO3′ bond angles at the residue connection sites. A double‐stranded helix is constructed from duplex building blocks (2 hydrogen‐bonded nucleotides) which are in turn built from the basic structural units. Every building block is constructed from two sets of geometric variables: {α, β, γ, χ, P, τ_max}, one for each strand. The building blocks are then assembled into a helix by using the 6 rigid body transformations {Δx, Δy, Δz, Θ_ROLL, Θ_TILT, Θ_TWIST}. For cartesian space programs, generating starting coordinates by this procedure is particularly useful as an alternative to using actual crystal structure coordinates. After describing the algorithm in detail, we illustrate how it was used to generate model A, B, and Z DNA helices. We conclude by suggesting how the algorithm can be used to pursue a build‐up technique and to set up a wide range of starting conformations in the goal of locating novel helical structures.