Missing Native Interface. Expected 6 Interfaces, Found 5

[luwei0917]

Thank you for providing this great program. We’ve encountered an issue and couldn’t determine the cause. Both our native and model proteins contain four chains, so we expect six interfaces to be output. However, we only received five. Do you have any insight into what might be causing this? DockQ A_protein_complex.pdb B_protein_complex.pdb --allowed_mismatches 80

two_pdbs.zip

Thanks!

[UTH-Arnav]

I have encountered this issue when running on python as well

from DockQ.DockQ import load_PDB, run_on_all_native_interfaces
native = load_PDB("8vef.pdb")
model = load_PDB("model.cif")
chain_map = { 
            "H": "H",
            "L": "L",
            "A": "T"
        }
res = run_on_all_native_interfaces(model, native, chain_map=chain_map)
print(len(res[0])) # Got 2 instead of 3, "HL" and "HA" predicted but no "LA"

pdbs.zip

I suspected it was due to missing residues, but I've successfully run DockQ on other files that have missing residues so that does not seem to be the case. I haven't figured it out yet. Any ideas?

[UTH-Arnav]

I was able to find a quick fix for this problem, though this will only work with a local repo so you will need to install DockQ again.
Run git clone https://github.com/bjornwallner/DockQ.git.
Navigate to the file DockQ/src/DockQ/DockQ.py and comment out lines 291 to 294. These lines contain an if statement that checks if there is no native interface between the selected two chains. In that case it returns nonetype and therefore that selected interface does not get calculated and terminates without any output. Bypassing this if statement forces DockQ to look at all chain combinations.
After that change, run pip install .
Let me know if that worked.

[UTH-Arnav]

In some cases, my fix can cause a division by zero error so you might have to make another change in the same file at line 425. Change the f1 function to

def f1(tp, fp, p):
    if tp + fp + p == 0:
        return 0
    return 2 * tp / (tp + fp + p)

then run pip install . again