HADDOCK score, complex bind energy

HADDOCK
1

Hello
How does HADDOCK calculate the Gibbs energy of interaction before and after protein complex binding (algorithm)?

2

Dear Nafise,

In HADDOCK, there is not computation such as the Gibbs free energy.

We do calculate a score (named HADDOCK score), which is a sum of weighted terms:

HADDOCKscore = 1.0 × E_vdw + 0.2 × E_elec + 1.0 × E_desol + 0.1 × E_air

This score allows us to rank/sort generated models, but not to predict the strength of the interaction, nor the binding affinity.

Other tools, such as Prodigy, are more dedicated to that purpose.

Cheers

1 Like
3

Thank you dear friend

در تاریخ سه‌شنبه ۱۲ نوامبر ۲۰۲۴،‏ ۱۱:۳۰ Victor Reys via BioExcel <notifications@bioexcel.discoursemail.com> نوشت:

4

Hello dear,
Can we use HADDOCK results to create mutations in the amino acid sequence? (so that in one amino acid, the lowest number obtained in HADDOCK score after replacing with different amino acids is chosen as the best mutation created).

5

Recent unpublished work we did showed that there is indeed a 0.35 Pearson correlation in predicting the mutation effect using the HADDOCK score. This was evaluated on the SKEMPI dataset.
For this, you can:

  • Must use the exact same conformation
  • Edit by hand the residue you want to modify in the PDB. (change the residue 3 letter code to an other one for all the ATOM records of that residue. e.g: SER → LYS, for N, C, CA, CB, O (at least the backbone atoms and the CB)
  • submit it to the refinement webserver (HADDOCK Refinement Interface), where no rigidbody docking and only the refinement of the interface will be performed.
  • Use also external tools to guide your mutation search (e.g: with PROT-ON http://proton.tools.ibg.edu.tr:8001/)
1 Like
6

Small clarification: you must change the residue name of all atoms in the residue. No need to delete or add the missing atoms. haddock will do it for you.

1 Like
7

Thank you for the clarification.

در تاریخ چهارشنبه ۱۳ نوامبر ۲۰۲۴،‏ ۱۲:۰۳ Alexandre Bonvin via BioExcel <notifications@bioexcel.discoursemail.com> نوشت:

8

Hello dear
I have two protein complexes, one containing 605 amino acids (structure A) and the other containing 291 amino acids (structure B). In the HADDOCK score, structure A is approximately -119 and structure B is -87. In both complexes, 170 amino acids, which are related to one protein of the complex, are similar, and the other protein is different. I evaluated these two complexes with PRODIGY, and structures A (-11.1) and B (-10.1) were obtained. I intend to show the structure of complex B better How should I have a proper justification or what method can I use?

9

Do realise that HADDOCK scores are not corresponding to binding affinity.

Similarly, PRODIGY is a model and as such perfect.
To get some ideas of how stable are the PRODIGY predictions you could repeat those with several members of the best HADDOCK cluster.

1 Like