Incorporating unspecific interaction restraints together with specific ones in HADDOCK Modeling

HADDOCK
1

Issue with Incorporating NMR Data in HADDOCK Modeling along with backbone interactions based on homologous, already determined structures

I am currently working on a university project where I need to model the Sxl-Unr-msl2 mRNA ternary complex in HADDOCK. The modeling is based on information from homologous proteins, NMR data, PREs, SAXS, and other relevant data sources. To facilitate this process, I developed a program that generates a TBL file containing distance restraints between backbone atoms, derived from homologous structures available in the PDB.

However, I am encountering a significant issue: the additional restraints from NMR data are not being considered during the modeling process. I created the additional TBL file lines using GenTBL and the restraints are of course rather unspecific compared to the previous assignments. I have attempted to force their inclusion by adding these restraints multiple times to the TBL file and by specifying them separately as unambiguous restraints, but these approaches have not yielded the desired results.

I would appreciate any advice or suggestions on how to ensure that the additional NMR data are incorporated effectively in the modeling process. Specifically, I would like to understand how to enforce the inclusion of these data within HADDOCK’s framework.

2

If you NMR restraints are defined to protons, you need to switch the option to keep all hydrogens.
By default the non-polar hydrogens are removed

3

My NMR restraints are not atom-specific. I am seeing CSPs in an alpha helix that is not involved in any interaction yet. The RNA 3’-end is currently unbound and already in proximity to the alpha helix (but yet too far away for a real interaction). So I created AIRs for the RNA 3’-end and the residues of this helix, only considering residues and no specific atoms, for example:

assign ( resid 14  and segid P)
       (
        ( resid 137  and segid A)
     or
        ( resid 138  and segid A)
     or
        ( resid 139  and segid A)
     or
        ( resid 140  and segid A)
     or
        ( resid 141  and segid A)
     or
        ( resid 142  and segid A)
       )  2.0 2.0 0.0

There are just four assignments for this interaction. Compared to that, my homology-based restraints are atom-specific and there are a lot of these restraints, for example:

assign (segid P and resid 12 and name N1) (segid X and resid 60 and name N) 7.27 2.0 2.0

A is Sxl, X is Unr, P is the RNA.

4

The question is: Are they not considered or are they not satisfied? These are different things.

To answer this question you would need to look at an output file of the docking and search for error indicating some restraints were not used because of wrong atom selection.

Did you use the web server? Can you point me to a run?

5

So, I am sure they are not satisfied, but considered, and I am sorry for the confusion. It is probably also an issue that my restraints violation energy is already high. An exemplary run is ID 382246. Is there some way I can “force” these restraints to be satisfied in case I am sure that they should be met?

6

You can always provide them as unambiguous-restraints.

Note that in standard ambiguous-restraints, we proceed to a random removal of 50% of them, to sample an increased set of conformations, and possibly remove noise from them.

1 Like
7

Hi there

I took a look at your run. A few things

  1. Your restraints are defined in ambig file, which means by default 50% will be randomly deleted for each model generated. If you want to use them all, rather upload them as unambiguous restraints

  2. Some restraints are not taken into account because of non-existing atom (e.g. the 1st base has no phosphate group by default) or wrong naming. Check one of the generated PDB for the atom naming of the RNA. The phosphate oxygens should be O1P and O2P instead of OP1 and OP2. Look for example in the complex_run1_it0_refine_1.out file and search for the following string: NOE>assign - you will see that for some distance restraints no atoms are selected.

And of course if large conformational changes are required to accommodate the restraints, then the protocol should be adapted, giving more flexibility to some parts. And even then, large conformational changes are a challenge!

PS: OP1/OP2 seems to be the official naming of the phosphate oxygens in the PDB, we should correct that on our side (I thought we did in the past)

1 Like