protein-DNA binding

HADDOCK
1

Dear Developer,

I am conducting research on protein-DNA binding, with a specific goal of designing novel proteins that can bind to designated DNA sequences from scratch.

  1. I am currently using HADDOCK 2.4 for protein-DNA docking. However, since the proteins I’m working with are entirely new, I have been selecting binding sites somewhat randomly by choosing a few residues. Sometimes, this results in significant constraint violations. Will this affect the credibility of the final results?
  2. I would also like to understand at what HADDOCK score and energy levels it would be reasonable to consider experimental validation.

I am very much looking forward to your response. Thank you so much for your assistance.

2

Additionally, due to permission limitations, all my other parameters are set to HADDOCK’s default values.

3
  1. I am currently using HADDOCK 2.4 for protein-DNA docking. However, since the proteins I’m working with are entirely new, I have been selecting binding sites somewhat randomly by choosing a few residues. Sometimes, this results in significant constraint violations. Will this affect the credibility of the final results?

Randomly choosing residues for a design purpose seems a rather bad idea…

You can always remove the restraint energy from the HADDOCK score (simply subtract 10% of the restraint energy from the score)>

  1. I would also like to understand at what HADDOCK score and energy levels it would be reasonable to consider experimental validation.

There are no such levels that can be defined as this changes per complex…
And score is NOT binding affinity.

4

Dear Developer,

In my protein design process, the lack of a specific binding site has resulted in some residue selections being inaccurate. If the final score, after applying a 10% penalty, is close to that of an experimentally validated protein structure known to bind the target DNA, can it be inferred that this protein is likely to bind with the target DNA?

Thank you for your assistance!

Alexandre Bonvin via BioExcel <notifications@bioexcel.discoursemail.com> 于2024年11月4日周一 03:54写道：

5

This was never really tested on our side…

And score is not binding affinity. So unless you have a benchmark to test your method it will be hard to make any reliable predictions.

6

Hello!

Regarding the prediction of protein-DNA binding, can the results from HADDOCK be considered as indicating a relatively probable binding conformation? Or do you have any better suggestions? Your advice would be incredibly valuable to me. Thank you for your help and patience in responding!

Alexandre Bonvin via BioExcel <notifications@bioexcel.discoursemail.com> 于2024年11月4日周一 15:55写道：

7

I would say “no”, but:

  • provided with perfect restraints (like the ones found in the crystal structure), you will reach close to 100% success rate.
  • provided with good restraints, it will drop to ~50-90% success rate
  • in ab-initio mode, only around 10% success rate
  • provided with wrong restraints, you can easily reach 0% success rate.

Everything depends on the quality of the restraints you are providing.
If you trust your restraints, you can trust HADDOCK predictions.

HADDOCK (and other docking tools) will always produce a binding conformation, but remember that they are only predictions that should be validated.

8

Hello,

So, can I consider the quality of these restraints as the strength of binding energy?