unstable tube orientation in morph

[Elaine Meng]

When I play back the morph in attached session with

coordset #3 1,21 pause 5

… you can see the yellow helix barely moves at all. However, when I show it as a tube, it bucks up and down (and so do some of the other helices):

cartoon style #3 modeh tube
coordset #3 1,21 pause 5

Assuming the session will stop working, how I made it was something like the following commands, which I will also attach as a cxc file:
open 1tag
open 1tnd
mm #2/a to #1/a
del /b,c
morph #1,​2
color bymodel
color byhet
hide #1,​2 targ ac
show #2&ligand
color #3:167-172 yellow targ c

[Elaine Meng]

The tubes also get fatter and thinner, another reason for needing a radius option as requested in #585

[Elaine Meng]

maybe the orientation instability is the same or similar to that with short strands #589

[Conrad Huang]

Fixed in 5f5b11940.

The unstable helix is six residues long. The helical axis was being computed using principal axis and that was generating poor initial estimates for some coordinate sets. Previously, axes for helices five resides or shorter were computed by matching against an ideal alpha helix. Extending the ideal helix to nine residues solved this problem by generating better initial axes estimates. Hopefully, for helices longer than nine residues, the principal axis estimates are sufficient because they should be much longer than wide and therefore provide reasonable estimates.