AMBER Archive (2004)

Subject: Re: AMBER: mutation of Tyr to Leu with TI

From: Thomas Steinbrecher (
Date: Wed Aug 25 2004 - 02:17:34 CDT

Deal Oliver

> is it possible to "mutate" a Tyr to a Leu within one thermodynamic
> integration simulation with sander?
> My understanding is that this would involve the following:
> o conversion of the OH in the Tyr sidechain and the carbon atom bound to
> it to dummies,
> o conversion of the CE carbons to hydrogens (HC) and of the H's bound to
> them to dummies,
> o conversion of the CD bound H's from HA to HC,
> o conversion of the CD carbon types from CA to CT,

all of this should be possible.

> o letting a third hydrogen (HC) appear at each of the CD carbons.

but here it gets difficult.

sander is not equipped to deal with appearing atoms at the moment. To
prevent the 'origin singularity effect' of vdW parameters reaching or
starting at zero, you need to employ a different mixing rule than linear
mixing in your TI potential function. In sander there is only a modified
mixing rule for disappearing atoms, not for appearing. The amber8 manual
p. 132ff gives a short introduction to this, but you might want to look
also at:

Simonson T., 1993 Mol. Physics 80 2, p. 441

I have tried to modify sander so that appearing atoms would be possible,
but this is not working yet. Right now you probably have to split
your transformation in two processes with disappearing atoms, first
something like Tyr->Ala then Leu->Ala and summing up both deltaGs.

It might also be a good idea to further split up your simulations in one
charge changing step (every atom gets the charge it would have in the
final state, eg 0.000 for dummies) and another step changing atom types
etc. because I had troubles dummying an atom while it simulataneously lost
its charge.

Drop me a mail if you are interested in my modifications, I will tell you
when I got them working.

Kind Regards,


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to