Pmx - problems reproducing benchmarking from de Groot 2020

pmx
1

Hi All,
I’m trying to reproduce some of the TI benchmarking done in the following paper:
“Large scale relative protein ligand binding affinities using non-equilibrium alchemy”

using pmx and the data provided in the GiHub repository referred to in the paper.

In order to produce the hybrid topology I did the following:

Step 1:
python one_ff_file.py -ffitp lig_CHEMBL3402752_30000_12/ffMOL.itp lig_CHEMBL3402748_5300_8/ffMOL.itp -ffitp_out ffMOL.itp

Step 2:
python atoms_to_morph.py -i1 lig_CHEMBL3402752_30000_12/mol_gmx.pdb -i2 lig_CHEMBL3402748_5300_8/mol_gmx.pdb -o1 pairs.dat -alignment -H2H

Step 3:
python make_hybrid.py -l1 lig_CHEMBL3402752_30000_12/mol_gmx.pdb -l2 lig_CHEMBL3402748_5300_8/mol_gmx.pdb -itp1 lig_CHEMBL3402752_30000_12/MOL.itp -itp2 lig_CHEMBL3402748_5300_8/MOL.itp -pairs pairs.dat -oa merged.pdb -oitp MOL.itp -ffitp ffmerged.itp -scDUMd 0.1

Step 4:
python one_ff_file.py -ffitp ffMOL.itp ffmerged.itp -ffitp_out ffMOL.itp

However, I noticed that in step 2, the pairs.dat file that is being produced is different than the one provided with the benchmarking data. Specifically, the mapping of the atoms between the two ligands is different.

All the input files and scripts I used are from the pmx GitHub repository and I made no changes.
I run pmx on a mac within a conda environment with python2.7.

Any suggestions on what could be the reason for these differences?

The original pairs file can be found here:

The ligands pdb can be found at:
pmx/protLig_benchmark/cmet/ligands_gaff2/lig_CHEMBL3402748_5300_8 at master · deGrootLab/pmx · GitHub and lig_CHEMBL3402752_30000_12

And my pairs.dat file looks like this:

1 1
2 2
3 6
4 5
5 4
6 3
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 20
25 22
27 21
28 28
32 33
33 34
34 35
35 36
36 37
37 38
38 39
39 40
43 44

Any help would be much appreciated!
Cheers,
Gal.

2

The difference comes from using -alignment flag. I suggest trying to run the command without this flag: this way both MCS and alignment pathways in the script will be probed. The mapped atom pairs then should be the equivalent to those deposited on the pmx github repository (the ordering of the mapped pairs may be different, but it has no influence on the further steps).

3

Thanks Vytautas!
I did as you suggested and it worked like a charm.

Cheers,
Gal.