Different hotspots are predicted by SpotOn when the chain IDs are reversed

I have noticed that when I switch the chains A and B in the submission form, the predicted results change. For instance, please consider this example -

I submitted a pdb file for PDB structure 1A22

When I submitted Chain A as interface 1 and Chain B as interface 2, the hotspots predicted were:
|spots.PDBResNo|spots.PDBResName|spots.PDBChain|
|21|HIS|A|
|168|LYS|A|
|271|ARG|B|
|304|TRP|B|
|369|TRP|B|

When I submitted Chain B as interface 1 and Chain A as interface 2, the hotspots predicted were:
|18|HIS|A|
|322|CYS|B|
|327|GLU|B|
|369|TRP|B|

From what I understand, theoretically, the interfaces remain the same and so the algorithm should give me the same results when I switch the chains. Am I missing something here, or is this a feature/limitation?

It looks like you have discovered a feature…

We are revisiting the code in order to modernise it.

In the mean time the only advice is to define the interfaces in the order they appear in the PDB file.

@amjjbonvin Thank You for your reply :slight_smile:

There’s another thing I’d like to add. I changed the names of the chains from A, B to X,Y respectively in pymol. I again ran SPOTON on them and the results are not the same as the ones I got before -

When I submitted Chain X(formerly A) as interface 1 and Chain Y(formerly B) as interface 2, the hotspots predicted were:
|spots.PDBResNo|spots.PDBResName|spots.PDBChain|Earlier Chain ID|
|18|HIS|X|A|
|21|HIS|X|A|
|45|LEU|X|A|
|175|THR|X|A|
|271|ARG|Y|B|
|304|TRP|Y|B|
|322|CYS|Y|B|
|369|TRP|Y|B|
|418|ASN|Y|B|

When I submitted Chain Y as interface 1 and Chain X as interface 2, the hotspots predicted were:

|spots.PDBResNo|spots.PDBResName|spots.PDBChain|Earlier Chain ID|
|18|HIS|X|A|
|304|TRP|Y|B|
|322|CYS|Y|B|
|369|TRP|Y|B|

The coordinates of the file are the same, I only changed their chain names.

Seems the same as before:define the interfaces / molecules in the order they appear in the PDB file

Okay, Thank you