Choose between motif search (global-semiglobal), global (Needleman-Wunsch) and local (Smith-Waterman) alignment.

In each alignment algorithm, similarity depends on dihedral/pseudodihedral angles, the nucleotide identities (A,C,G,U), and the base-pairing in both query and target.

Motif search involves a global alignment of the query and a semiglobal alignment of the target. Suboptimal alignments may be produced by entering a nonzero real parameter X in the interval [0,1] in the blank for "suboptimal alignments score ratio". If S denotes the similarity score for the optimal alignment, then suboptimal alignments having score T will be displayed in a pull-down tab in the upper right corner of the output screen, for all values T such that (S-T)/S ≤ X. Useful typical values for the suboptimal alignment score ration are 0.1, 0.2, 0.3.

The user may enter nonnegative parameters for
*dihedral*,
*nucleotide sequence* and
*base pairing* similarity. If parameters are respectively x,y,z,
then `DIAL` applies a weight of
`x/(x+y+z)` for dihedral/pseudodihedral angle similarity,
`y/(x+y+z)` for nucleotide sequence similarity and
`z/(x+y+z)` for base pairing similarity.

Similarly, the user may enter nonnegative parameters for
dihedral angles
α, β, γ, δ, ε, ζ, χ and
pseudo-dihedral angles
η, θ. If dihedral parameters are respectively
x1,...,x7 and pseudo-dihedral parameters are y,z, then `DIAL`
applies a weight of
`x1/(x1+...+x7+y+z)` for dihedral angle α similarity, etc.
and `y/(x1+...+x7+y+z)` for pseudo-dihedral angle η similarity.

There are correlations between various dihedral angles. We have found useful the following parameter settings:

- χ,η,θ set to be 1, all other dihedral angles set to 0,
- η,θ set to be 1, all dihedral angles set to 0.

Default gap costs and nucleotide match and mismatch values are taken from BLAST. Our ROC curve analysis of the SCOR database suggests that a good parameter choice to weigh dihedral/pseudo-dihedral angle, nucleotide and base pairing similarity is respectively 0.8, 0.2, 1.0 or similar values. The effect is to emphasize the contribution of dihedral/pseudo-dihedral angles over that of nucleotide sequence similarity, while weighing the contribution of base pairing nature equally with the combined contribution of dihedral/pseudo-dihedral angles and sequence.