On Thu, Nov 14, 2013 at 8:32 AM, Davide Baù <davide.bau@gmail.com> wrote:

Hi,

I got both the MD and the BD to work, but there is something I don't understand about the wlc.

If I set it up and like this (with both MD and BD):
l_max = 100.0
lp = 5.0
wlc = IMP.misc.WormLikeChain(l_max, lp)
sd = IMP.core.SphereDistancePairScore(wlc)

crw = IMP.container.PairsRestraint(sd, cpc)
crw.set_weight(kforce)
m.add_restraint(crw)

And then check the l_max and lp values with:

sl_max = wlc.evaluate_with_derivative(l_max)
slp = wlc.evaluate_with_derivative(lp)
print "l_max %f\nlp %f" % (sl_max[0], slp[0])

I get "out of scale" values, even if wlc is the only score applied, e.g.:

l_max 590.102905
lp 0.022517

Another thing that I noticed is that, apparently, only l_max has a influence on the final structure (it constrains the occupancy). Is that expected?

The WLC score has a hard maximum on its max length (and goes to infinity as you approach it) and so values above that will be out of bounds. As such, you have to be pretty careful when using it so as not to have things blow up. Also, using WLC with sphere distance pair score seems a bit odd to me, but I'm not a polymer physicist :-)

In addition, it is pretty flat away from the maximum, so other than that it won't, as you are seeing, have much effect on the structure. Sorry about the slow response.