Skip to content

Latest commit

 

History

History
46 lines (45 loc) · 2.96 KB

algorithm.md

File metadata and controls

46 lines (45 loc) · 2.96 KB
Raw

energetics_daoe - new

  • Consider two ends with some offset. The "first" end of the two will be read 5 to 3, the second 3 to 5 -> "left to right" in a diagram.
  • For each paired pair of bases (2 nt on one end, 2 on the other), calculate three values:
    • ens: if a matching NN pair, then the dG for the pair, otherwise 0.
    • ltmm: if only the "righthand" bases match, then the dG for a terminal mismatch on the left-hand side (5' on the first end, 3' on the second), otherwise 0. Note that currently this is just the sum of two terminal dangle terms.
    • rtmm: if only the "lefthand" bases match, then the dG for a terminal mismatch on the right-hand side, otherwise 0.
    • intmm: in either of the previous two cases, the dG assuming the mismatch is internal, otherwise 0.
  • Depending upon the offset:
    • For zero offset, if they are already matching, ens[0] and ens[-1] have the dG for adjacent coaxial stacks added to them (optionally with nick correction), and have the dG for the preexisting dangles in the unbound ends subtracted. This gives adjusted binding dGs for those bases. Note that this ignores an unusual case where the end of one of the ends has an internal mismatch. It's unclear how realistic our parameters would be in that case, or whether it would ever reasonably be favorable.
    • For "stretching" offset, ens[0] and ens[-1] have terminal dangle parameters added to them for the continuation of the ends.
    • For "contracting" offset, ens[0] and ens[-1] have the adjacent coaxial stacks added, and tail penalties.
  • Now, reading left to right, by NN pairs, we start with an accumulator (acc) of 0, and a max binding strength (bindmax) of 0:
    • If the pair is matching, then ens is added to the accumulator.
    • Otherwise, if the pair can be a right terminal mismatch (rtmm != 0), then if bindmax < acc + rtmm, then bindmax is set to acc + rtmm, which corresponds to the end of binding with a dangle. Then intmm is added to the accumulator, corresponding to the start of an internal mismatch.
    • Otherwise, if the pair can be a left terminal mismatch (ltmm != 0), and we're not on the last pair, we check to see whether ltmm > acc+intmm and the next pair is matching, in which case we assume we have a left terminal mismatch, as it will be stronger, and we set (not add) acc = ltmm; otherwise, we add intmm to acc (we assume an internal mismatch). Note that this ignores the possibility of just the last base being paired, with the penultimate an internal mismatch. This is unlikely and Santa Lucia advises against using internal mismatch parameters in such cases.
    • If none of these are the case, add a "loop penalty" to acc, which approximates the dG penalty for a symmetric loop.
  • At the end of this process, if acc > bindmax, set bindmax = acc.
  • Do this for every offset, and find the maximum bindmax.