Positive energy in it0 with fix_origin_mol

HADDOCK
1

Hi @amjjbonvin,
Does fixing the origin of molecules (fix_origin_mol1=true) affect the energies in the system, especially “vdw”? I am doing different levels of docking, in each one, I put all proteins from the previous dock as one PDB file and freeze them and do docking of new proteins on the previous ones. Now that the system is growing really large, the scores in the “file.list” file are becoming positive. I just want to know if getting a positive score is still meaning full.

As an example, I have four PDB files fixed, each containing multiple proteins.
I use only it0 and set the “structures_0=200”. The lowest score I get out of 200 structures is: 93195.851

Here is the relevant parts of the pdb output file with lowest energy:

REMARK ===============================================================
REMARK            total,bonds,angles,improper,dihe,vdw,elec,air,cdih,coup,rdcs,vean,dani,xpcs,rg
REMARK energies: 9.693252E+06, 0, 0, 0, 0, 9.641654E+06, -3502.86, 55101.3, 0, 0, 0, 0, 0, 0, 0
REMARK ===============================================================
REMARK            bonds,angles,impropers,dihe,air,cdih,coup,rdcs,vean,dani,xpcs
REMARK rms-dev.: 0,0,0,0,95.0777,0,0, 0, 0, 0, 0
REMARK ===============================================================
REMARK               air,cdih,coup,rdcs,vean,dani,xpcs
REMARK               >0.3,>5,>1,>0,>5,>0.2,>0.2
REMARK violations.: 44, 0, 0, 0, 0, 0, 0
REMARK ===============================================================
REMARK                        CVpartition#,violations,rms
REMARK AIRs cross-validation: 2, 0, NaN
REMARK ===============================================================
REMARK NCS energy: 0
REMARK ===============================================================
REMARK Symmetry energy: 0
REMARK ===============================================================
REMARK Membrane restraining energy: 0
REMARK ===============================================================
REMARK Local cross-correlation:  0.0000
REMARK ===============================================================
REMARK Desolvation energy: 112
REMARK Internal energy free molecules: -6.31537
REMARK Internal energy complex: -6.31537
REMARK Binding energy: 9.638263E+06
REMARK ===============================================================
REMARK buried surface area: 38084.2
REMARK ===============================================================
REMARK water - chain_1: 0 0 0
REMARK water - chain_2: 0 0 0
REMARK water - chain_3: 0 0 0
REMARK water - chain_4: 0 0 0
REMARK water - chain_5: 0 0 0
REMARK water - chain_6: 0 0 0
REMARK water - chain_7: 0 0 0
REMARK water - chain_8: 0 0 0
REMARK water - chain_9: 0 0 0
REMARK ===============================================================
REMARK water - water: 0 0 0
REMARK ===============================================================

Is what I do still safe? :slight_smile:

Best,
Mohammad

2

The positive Evdw energies (which translate also into positive scores) come from clashes in your structures.

Also if you fix several molecules, their interaction energy is still being calculated and part of the scoring.
A better solution could be to shift the residue numbering in that case and make those a single molecule.

Note also that the it0 models are not refined, hence the high energies.
Basically you are only doing rigid body docking…

3

Thanks @amjjbonvin

this is what I originally do after each dock:

I run a short “standard MD”, including solvation, adding ions, energy minimization (10,000 steps), equilibration (NVT 2ns, NPT 0.2ps) and finally, a 5ns NPTproduction run, to avoid any clashes. During the simulation, the system is stable and I don’t see any “explosions” due to the clashes and the high energy.

I put the structures after the MD in a single PDB, and renumber residues from 1 to the end.
I go from 1 to around 7400 in one file if there are more I go with two PDBs.
There are no “TER” between the proteins in a PDB to treat the whole as one molecule, just one “TER” before the “END” and freeze the PDB and use this PDB to dock the next proteins.

and by doing all these, I get the high van der Waals energies (I have them in my previous message).

am I missing a point?

4

Sounds to me it would be way simpler to run the it1 stage of HADDOCK…

5

I agree :slight_smile: @amjjbonvin
I guess I forgot to mention that I do CG docking.
The problem is that I have a ligand that is not defined in the haddock, so the docking crashes when it goes from it0 to it1. I have defined it with simple beads (described in the literature) in the param files
So, instead of trying to hack the HADDOCK further for it1, I decided to go around it with external MD simulations.

Nevertheless, do you agree that what I do is theoretically meaningful? Am I doing something that could cause the large Evdw?

6

Are the systems that large that you need CG?

And if the vdw energies are that large something is wrong somewhere…

7

The system will have more than 100 proteins (currently has ~ 40) with residues ranging from ~700 to 1,300. Trying to see the maximum I could reach with HADDOCK

for the EvdW, I can not think of structure clashes as the external MD runs fine

hmm, one last point that I could think of now would be the changes I make to the cns file:
I make the following changes in my run.cns file:

sed -i 's/fix_origin_mol1=false/fix_origin_mol1=true/g' run1/run.cns
sed -i 's/fix_origin_mol2=false/fix_origin_mol2=true/g' run1/run.cns
sed -i 's/fix_origin_mol3=false/fix_origin_mol3=true/g' run1/run.cns
sed -i 's/fix_origin_mol4=false/fix_origin_mol4=true/g' run1/run.cns
sed -i 's/fix_origin_mol5=false/fix_origin_mol5=true/g' run1/run.cns

sed -i 's/cmrest=false/cmrest=true/g' run1/run.cns
sed -i 's/kcont=1.0/kcont=0.1/g' run1/run.cns

sed -i 's/epsilon_0=10.0/epsilon_0=1.0/g' run1/run.cns
sed -i 's/dielec_0=rdie/dielec_0=cdie/g' run1/run.cns
sed -i 's/dielec_1=rdie/dielec_1=cdie/g' run1/run.cns

sed -i 's/structures_0=1000/structures_0=200/g' run1/run.cns
sed -i 's/structures_1=200/structures_1=0/g' run1/run.cns
sed -i 's/anastruc_1=200/anastruc_1=0/g' run1/run.cns

am I missing anything? should I change other stuff?

8

@amjjbonvin

9

In short nothing wrong with your settings…

10

Thank you. I appreciate it.