Haddock3 perform multi-protein docking

HADDOCK
1

Hi, I conducted multi-protein docking on my local linux, and there were four protein.
1st step: I used Alphafold 3 to generated a protein complex, and output an ambig file
2nd step: I token AF3_complex.pdb as reference pdb，and ambig.tbl as restraint file to perform four protein molecules docking
Here is my cfg script：

# ====================================================================
# 4-protein docking with HADDOCK3, no restraints, no His auto-processing
# ====================================================================

run_dir     = "run1-full"
mode        = "local"
debug       = true
concat      = 3
queue_limit = 100

# ====================================================================
# Paths to the four single-chain protein PDB files
# (names: proteinA, proteinB, proteinC, proteinD)
molecules = [
    "pdbs/proteinA.pdb",  # Contains only chain A
    "pdbs/proteinB.pdb",  # Contains only chain B
    "pdbs/proteinC.pdb",  # Contains only chain C
    "pdbs/proteinD.pdb",  # Contains only chain D
]

# ====================================================================
[topoaa]
autohis = false
# If you have special histidine protonation states (e.g. HIE/HID), 
# you can add nhisd/nhise per molecule; otherwise, leave this as is.

# ====================================================================
[rigidbody]
tolerance   = 5
sampling    = 500
# No ambig_fname specified → unrestrained free docking

# ====================================================================
[caprieval]
reference_fname = "pdbs/ref_complex.pdb"
# Reference structure for CAPRI evaluation

# ====================================================================
[seletop]
select = 100
# Select the top 100 models after scoring

# ====================================================================
[flexref]
tolerance = 5
# No restraints (ambig_fname not specified)

# ====================================================================
[emref]
tolerance = 5
# No restraints (ambig_fname not specified)

# ====================================================================
[clustfcc]
min_population = 1
# Minimum cluster size for clustering

[seletopclusts]
top_models = 5
# Select top 5 models from the best-scoring clusters

# ====================================================================
[caprieval]
reference_fname = "pdbs/ref_complex.pdb"
# Reference complex structure used again for final CAPRI evaluation
# ====================================================================

3rd step：pymol visulization

image
image1865×1366 90.3 KB

There is no sign of any docking
Can you help me solve this problem and optimize the script？
I’m very very grateful!

2

Hi there,
There’s no signs of docking because there are no restraints in your cfg to drive the docking.
You have to either supply ambig_fname to sampling and refinement modules, or opt for an ab initio docking. Check user manual for more info: Ab-initio docking mode - HADDOCK3 User Manual

3

There are no restraints at all defined for your docking…. Nothing will happen.

What do you want to do exactly?

Also in local mode, I would only specify the number of cores to use, queue_limit and concat are irrelevant.

4

I want to dock four protein molecules， but there are only AF3_Predict_complex.pdb and 4 protein.pdb

5

Then you need to turn on some of the ab-initio mode of HADDOCK.

For the rigibody module this mean adding:

cmrest = true

and for the flexref and emref steps:

contactairs = true

And you will need to greatly increase the sampling, to e.g. at least 10000 for rigid body and select 400 to 1000 for the refinement stages.

Even then with this number of molecules and no info at all, I would not be very confident you will get anything that makes sense…

6

Hi！
Then Would it be better if I used haddock-restraints to generate restaints by AF3.pdb
Do you have any further suggestions? I‘ll appreciate for your recommendations

7

Then Would it be better if I used haddock-restraints to generate restaints by AF3.pdb
Do you have any further suggestions? I‘ll appreciate for your recommendations

If you want indeed to build the complex based on the AF3 interfaces that’s a good way to do it.

8

Thanks for your help

9

Thanks for your support