Hi Jose,
The modelling of the TMs themselves shouldn't be a problem, if you have a reliable template which is from the same family (e.g. GPCR). For instance, I'm modelling a GPCR using rhodopsin as a template and for each of the 7 TMs the sequence alignment is clear. That means the relative positioning (TM orientations) of the helices should be correct. Moreover, certain key residues that connect the helices are conserved within the GPCR family and I'm planning to add additional constraints to these interhelical distances in the modeller optimization.

As far as I can see, the modeller energy function doesn't contain any term that should be biased towards globular proteins such as hydrophobic/hydrophilic terms, solvent accessability term or the like. The only problem that I see is that the data for some parameters (e.g. side-chain conformation) is inherently different in globular and membrane proteins, due to packing. The statistical data used in the modeller energy function is solely based on globular proteins as far as I understand.
All other energy terms used in modeller seem to me to be general to all proteins.
I would like the confirmation about this from the modeller team, though, since this is only my uneducated understanding.

My main question was that the homology of GPCRs is relatively low, but the topology is better conserved than in globular proteins of similar homology. Thus the "greater than 40% homology rule" shouldn't apply to GPCRs.

I hope this helps,
Dan

José Saldanha wrote:

Hi dan,
I saw your message on the modeller list. Unfortunately I cannot
answer it, but have a question to you. Can you use modeller's
optimisation for TM regions? I have done it myself, but feel
uneasy since I imagine it is meant for globular proteins, not
TM regions.
regards,
-josé-
=================================================================
José W. Saldanha (PhD)            Tel: (+44)(0)208 913 8635
Division of Mathematical Biology Fax: (+44)(0)208 913 8545
National Institute                Ema: jsaldan@nimr.mrc.ac.uk
    for Medical Research (NIMR)   Cel: 07762 340964
Mill Hill, London NW7 1AA (UK)
                         Humanisation bY Design
                 http://www.cryst.bbk.ac.uk/~ubcg07s
=================================================================

-- 
Dan Thomas Major (at Dr. B. Fischer's lab)
Bar-Ilan University 
Ramat-Gan, Israel