Molecular docking can be an important step in analysing compound designs, filtering through huge chemical spaces and prioritising promising compounds.

In our Covid-19 project (check it out here and read the paper here), docking and binding energy calculations are a big part of Rosalind’s prioritisation phase. 

We know setting up a docking process for the scale necessary for ML requires time, expertise and generally considerable compute resources. We are committed to helping the community in these challenging times and we hope our collective effort will speed up the pace of innovation to deliver treatments to this disease. We are therefore offering the use of our computational docking pipeline to the wider community. 

System specifications:

  • We use Smina (a fork from Autodock Vina) for docking.
  • We use Gytsum-dl for compound preparation.
  • We run the process on CPUs and we can easily scale the process when and if necessary.
  • The speed of the process depends on the parameters (exhaustiveness, number of confirmations, size of the molecule, etc.). On average, we estimate an hour per 50 compounds.

How to use Rosalind’s docking capabilities:

The system is designed to dock compounds into the crystal structure of SARS CoV2 Mpro protease (PDB ID: 6YB7) and that of its closely related SARS CoV (PDB ID: 3V3M). We also included the recent crystal structure with PDB ID: 6w63. These are the ones we used in our paper and in a number of others. The easiest and fastest way would be to dock into one of these three targets as we already have the pdbqt and the configuration files prepared. Along with the target, you will need to upload the list of the compounds, obviously.

To execute the docking, head to our interface here and use one of the options to either test individual SMILES (if you only have a handful) or upload a txt file containing your SMILES (one per line).

By default, we will run the process with the following parameters: exhaustiveness: 8, num_modes (number of binding modes): 9 and a random seed.

If you would like to run the process using other configurations or if you have been working on other crystal structures than the ones above, either send us the PDB ID and we will prepare the files or send the prepared files (pdbqt and config) if you have them. You can send these via email to contact@glamorous.ai.

Once we have the data, you should expect an email with the docking results within 48 hours.  

Acknowledgement

We would like to thank Marawan Ahmed for his invaluable feedback.