
Hi! I'm using MODELLER to build a homology model of my protein. In my first try, I create a model without any restraint, and then I compared the secondary structure of this model (Using STRIDE and VMD) with the secondary structure predict by PSIPRED webserver. For the first 26 residues, I have these predictions: PSIPRED: CHHHHHHHHHHHHHHHHHHHHHHHHC Homology model: CCCCCCCCCCCCHHHHHHHHHHHHHC My idea is restraint the region where the predictions are coincidences (13:A to 26:A) and applying restraints to force an alpha Helix structure between residues 3 and 11. I didn't include the residues 2 and 12 in order to give some flexibility to MODELLER to connect the restraint to the remainder of the protein (see below). from modeller import * from modeller.automodel import * import sys log.verbose() env = environ() class MyModel(automodel): def select_atoms(self): return selection(self.residue_range('13:A', '26:A')) def special_restraints(self, aln): rsr = self.restraints at = self.atoms # Define an alpha helix rsr.add(secondary_structure.alpha(self.residue_range('3:A', '11:A'))) But, when I look this region on VMD, I don't have a Helix between residues 2 to 12. I don't know, what I'm doing wrong. Is a better approach to restraint as Helix from 2 to 26? Should I keep some regions without optimization during the restraint protocol?
Thank in advance,
Fred

On 4/14/25 3:49 PM, Fred Pontes via modeller_usage wrote: > Hi! I'm using MODELLER to build a homology model of my protein. In my > first try, I create a model without any restraint, and then I compared > the secondary structure of this model (Using STRIDE and VMD) with the > secondary structure predict by PSIPRED webserver. For the first 26 > residues, I have these predictions: PSIPRED: CHHHHHHHHHHHHHHHHHHHHHHHHC > Homology model: CCCCCCCCCCCCHHHHHHHHHHHHHC My idea is restraint the > region where the predictions are coincidences (13:A to 26:A) and > applying restraints to force an alpha Helix structure between residues 3 > and 11. I didn't include the residues 2 and 12 in order to give some > flexibility to MODELLER to connect the restraint to the remainder of the > protein (see below).
In general I would not recommend using secondary structure prediction when you have 3D structure (templates) available.
Modeller's secondary structure restraints are quite weak and are designed to 'nudge' the structure towards the desired conformation. They're not going to force a helix if the protein is far from helical to start with, and they certainly won't override any structure implied by the template(s). You didn't include your alignment here but if the region 13:A to 26:A is aligned with any template at any point, the template structure will likely win out over the secondary structure restraint.
Ben Webb, Modeller Caretaker
participants (2)
-
Fred Pontes
-
Modeller Caretaker