On Tue, 26 Aug 2003 15:20:16 +0200 Morgane Agez wrote:

> orientation of the diferent domains. How can I orient the three > template structures appropriatly before the modeling?

As it says in the manual - Modeller can automatically orient the domains if the template alignments are overlapping :

dom1:....... AAAAAA--------*

dom2:....... ----BBBBBBB---*

dom3:....... --------CCCCCC*

unk1:....... QQQQQQQQQQQQQQ*

This is an assumption - the orientation is given by the superposition of the C terminus of the preceding domain template to the N terminus of the following domain template (because they are aligned to the same parts of the model sequence). This assumption is not always correct- for a whole host of reasons, but its a good first approximation.

If you have no overlap, modeller can make no assumptions, and in fact it assumes that you have oriented the domains yourself. This means that it assumes that all the domains lie in the same 3D coordinate frame; so you can easily get knots, infeasible backbone connections, or far worse!

Solving the domain orientation problem is a little like macromolecular docking - except it has the additional constraint given by the connectivity of the domains through the alignment. There's not a reliable, automated method for solving this problem, as yet (afaik:-)).

If you do choose to attempt to orient the domains yourself, all you need is a package to manipulate and write out the transformed domain coordinates. Then its a matter of inspiration, and matching HP-HP patches on the surface of the domains, perhaps!

good luck.. j.

_______________________________________________________________________ Dr JB Procter:Biomolecular Modelling at ZBH - Center for Bioinformatics Hamburg http://www.zbh.uni-hamburg.de/mitarbeiter/procter