PostEra

Results from Nwat-MMGBSA workflow on covalent warheads

Hi guys,
I did some calculations on the covalent_warhead_db following the method described in DOI:10.3389/fchem.2018.00043

Briefly, the DB has been converted to 3D using MOE. The receptor was prepared strating from Mpro-x0830_0.pdb file. The structure was initially prepared using the “Structure Preparation->Correct” tool of MOE; the AMBER10:EHT ff and born solvation model were applied. Then, the covalent bond between Cys145 and chloroacetamide alpha carbon was broken and the system minimized by applying restraints to all heavy atoms, then by applying restraints on backbone atoms only. Then, the complex was protonated at ph=7.0 by Protonate 3D. The binding site Cys was protonated and I left it in this way.

The screening was done in a “non-covalent” manner, based on the assumption that the lowest will be the energy of the activated complex, the lowest will be the energy of the TS for covalent bonding (Hammond’s postulate)

The library was processed by requesting the enumeration of tautomers, generated by the software UNICON (only the most favoured tautomer was saved), stereoisomers and ring-conformations (generated by the software SPORES). Docking was done by PLANTS. The following “non default” parameters were added to the PLANTS input:

Binding site center=“9.50042 -1.03244 22.9953”
Binding site radius=14
speed=1
chemplp_protein_hb_constraint 2210 5
aco_sigma 7.0
Scoring function=ChemPLP

The receptor, in pdb format, is attached here cov2_protease.pdb (373.6 KB) .

The top 250 compounds were then rescored with the Nwat-MMGBSA method (see reference above) with Nwat=0 (standard MMGBSA), Nwat=30 (30 explicit waters surrounding the ligand) and Nwat=60 (60 explicit waters). Point charges for the 250 rescored compounds were computed with the AM1-BCC method.

All complexes were subjected to MD (pmemd.cuda), 1.63 ns of equilibration and 1 ns of production run, as described in the paper cited above.

The final scoring three additional booleans: DIST < 5 (distance-to-warhead of Cys145); RMSD LIG < 2 ( max RMSD of ligand during MD); ST.DEV < 10% (standard deviation of Nwat-MMGBSA energies for Nwat=30). Compounds respecting these three filters were then rescored by Nwat-MMGBSA energies for Nwat=30. Rescored compounds are available here: Rescoring_with_RMSD_and_DIST2.xlsx (36.9 KB)

Best,

Alessandro

I see that four compounds hit all four of your scoring criteria (ST.DEV < 10%, RMSD LIG < 2, DIST < 5, DIST < 3.5):

TAM-UNI-d1c-20
MAK-UNK-3c3-1
PAU-WEI-b9b-4
PAU-WEI-b9b-6

Interestingly, one of these compounds, PAU-WEI-b9b-4, was also voted a “top” compound by one of the community members who offered judgments. Five of the top 25 compounds in Alessandro’s list were also deemed top by one or more panelists, giving some credence to Alessandro’s model and/or the expertise of the panelists. In addition to the one mentioned, these include:

BOG-INS-177-1
MIH-UNI-e57-1
BOG-INS-d6b-1
BOG-INS-8eb-1

So, there are my votes! (note: I’m NOT qualified to vote, I’m just a data analyst who has noticed a pattern).

Also: I would happily pay to expedite the production of these eight compounds. Admins: please PM me if this is an option, as I know it’s something that has been considered.

UPDATE: I was not up on the latest. The following compounds have a status of “ordered” in the compound tracker:
PAU-WEI-b9b-4
PAU-WEI-b9b-6
BOG-INS-8eb-1

Still happy to pay for the other 5!

2 Likes

Great, thanks for your analysis! Just waiting on Weizmann to have their assay up and running (Oxford assay is currently full of compounds), and we will have a price!

Great! Consensus between different strategies is always a good thing to aid selection. Thank you for your analysis!

The SD-file with the full results is now uploaded to Fragalysis. You can download it from:
https://fragalysis.diamond.ac.uk/viewer/compound_set/AlessandroContini-NwatMM-GBSAcovalentrun1

1 Like

@matteoferla made an interactive view of the results on Michelangelo here: https://michelanglo.sgc.ox.ac.uk/data/fdb1b954-5d4c-405f-a081-4e042b81f892

Just as a further explanation on how to interpret rescoring data, my suggestion is to rank as:
DIST < 5 = Y (distance of electrophilic warhead carbon to CYS145 sulphur)
RMSD LIG < 2 = Y
ST.DEV < 10% = Y
MM-GBSA Nwat=30: the lowest (more negative) is the energy score, the better is the compound

My proposal is to select compounds respecting all the above “boolean” restrictions (30 compounds), prioritizing them by “MM-GBSA Nwat=30”. Among them, four compounds also respect a more restrictive DIST < 3.5 criterion; those compounds should also be prioritary.