swapaa may be using probability as dominant criterion

[Elaine Meng]

unless I am confused. Here is the example:

open 6pt2
del /B,D
hide targ ar
show :128 :<4
view clip false
swapaa :128 ASN

... chooses rotamer with 1 hb 1 clash (both calculated using default options) but the highest probability. (Sidechain angles: labels give chi1 -69.3 chi2 -19.3 slightly different from the -73,-19 in rotamer dialog) However, there are several rotamers with 0 clashes, some similar to the wild-type ASP position, that should be chosen over the one that is apparently used.

[Eric Pettersen]

Works for me in the latest daily build. Chooses the second most probable rotamer, which has 0 clashes.

I guess I need a recipe that actually reproduces the problem.

[goddard@…]

The magic ingredient to get the weird behavior is to open two structures, say two copies of of 6pt2 and then still use "swap :128 ASN" so it tries to swap both.  I guess it is detecting clashes between the overlapped structures.

In the delta opioid tutorial two structure 4n6h and 6pt2 are used and aligned with match maker and it is just sloppiness that the command used was "swap :128 ASN" instead of "swap #2:128 ASN".  On the other hand it would not be unreasonable to swap both at the same time, so I am not sure if treating the two structures as a complex is a good default.

[Elaine Meng]

Even if there are clashes with other models, seems like the rotamer-list dialog's "clashes" column is not listing them, so I'm not sure that that is the problem.

[Eric Pettersen]

I can reproduce the problem with two copies of 6pt2 open.

[Eric Pettersen]

You have to supply "ignoreOtherModels true" to the swapaa command. This works the same way as it did in Chimera.

You will have to get tomorrow's build, since there was a minor bug with "ignoreOtherModels true" (missed changing a variable name during the port from Chimera).

I am open to possibly changing the default, since the clashes/H-bonds dialogs in the tool version just automatically ignore other models. Or instead the tool version could give control over ignore/not ignore.

[Elaine Meng]

I vote for "ignoreOtherModels true" on by default.  But (less urgent) I would not mind if the dialogs opened from Rotamers exposed the option so that people could turn it off in the rarer cases in which they want it off.

[Eric Pettersen]

I have changed the default for 'ignoreOtherModels' from false to true.