Can the Modeller take periodic boundary conditions into account during modeling? I aim to construct a condensed model, but currently, unwarping the protein to the Modeller and then warping it back leads to incorrect overlaps of side chains. I noticed that symmetry constraints can be added, but my original system is already large, and adding additional images causes crashes.
On 4/3/24 12:56 PM, Qizan Chen via modeller_usage wrote: > Can the Modeller take periodic boundary conditions into account > during modeling?
No, Modeller has no built-in support for periodic boundary conditions.
You can add custom restraints to the scoring function (see, e.g. https://salilab.org/modeller/10.5/manual/node480.html) so in principle you could replace the existing restraints with PBC-aware versions, e.g. rather than calculating the distance between A and B, return the distance between A and the nearest image of B. However, this would be a significant undertaking.
> I noticed that symmetry constraints can be added
They can, but Modeller's symmetry restraints would likely not be very efficient here, since they don't remove any degrees of freedom.
> but my original system is already large, and adding additional images > causes crashes.
Modeller should not crash with any size system, unless you run out of physical memory, so I'm curious to know what your problem is here.
Ben Webb, Modeller Caretaker
It seems difficult to implement PBC with custom restraints for me, because I can't even don't know which default constraints need to be modified.
About the crashes, previously I ran on Ubuntu 18.04, it failed for segmentation fault. Today I updated my Linux to 22.04, and now it won't crash, but a max_molpdf issue arises.
Anyway, thank you so much for your reply!
Best regards! Qizan Chen
On Wed, Apr 3, 2024 at 5:09 PM Modeller Caretaker modeller-care@salilab.org wrote:
> On 4/3/24 12: 56 PM, Qizan Chen via modeller_usage wrote: > Can the > Modeller take periodic boundary conditions into account > during modeling? > No, Modeller has no built-in support for periodic boundary conditions. You > can add custom restraints > ZjQcmQRYFpfptBannerStart > This Message Is From an External Sender > This message came from outside your organization. > > ZjQcmQRYFpfptBannerEnd > > On 4/3/24 12:56 PM, Qizan Chen via modeller_usage wrote: > > Can the Modeller take periodic boundary conditions into account > > during modeling? > > No, Modeller has no built-in support for periodic boundary conditions. > > You can add custom restraints to the scoring function (see, e.g.https://urldefense.com/v3/__https://salilab.org/modeller/10.5/manual/node480...) so in principle > you could replace the existing restraints with PBC-aware versions, e.g. > rather than calculating the distance between A and B, return the > distance between A and the nearest image of B. However, this would be a > significant undertaking. > > > I noticed that symmetry constraints can be added > > They can, but Modeller's symmetry restraints would likely not be very > efficient here, since they don't remove any degrees of freedom. > > > but my original system is already large, and adding additional images > > causes crashes. > > Modeller should not crash with any size system, unless you run out of > physical memory, so I'm curious to know what your problem is here. > > Ben Webb, Modeller Caretaker > -- modeller-care@salilab.org https://urldefense.com/v3/__https://salilab.org/modeller/__;!!KwNVnqRv!FEsVB... > Modeller mail list: https://urldefense.com/v3/__https://salilab.org/mailman/listinfo/modeller_us... > >
On 4/4/24 2:30 PM, Qizan Chen wrote: > It seems difficult to implement PBC with custom restraints for me, > because I can't even don't know which default constraints need to be > modified.
Yes, you would need to parse the restraints file (https://salilab.org/modeller/10.5/manual/node503.html), replace some subset of the restraints with your custom restraints, and then rerun Modeller with that custom restraint file. As I said, it would be a lot of work!
> About the crashes, previously I ran on Ubuntu 18.04, it failed for > segmentation fault. Today I updated my Linux to 22.04, and now it won't > crash, but a max_molpdf issue arises.
Right, max_molpdf saves computer time by quickly discarding very bad models with large scores. But since max_molpdf is a fixed value (10^7) but pdf scales roughly with the square of the number of atoms in your system, for very large systems it might throw out valid models. So you can increase it in this case: https://salilab.org/modeller/10.5/manual/node52.html
Ben Webb, Modeller Caretaker