Welcome to the Ramirez Lab Wiki - Docking post-processing: Interaction frequency among multiple-cluster conformers
To analyze the frequency of receptor-ligand interactions on a set of poses from different clusters, we use scripts included on Schrödinger Suit (v.2020-3) to calculate the interactions on every pose and a in-house functional workflow built on Knime to get the frequency of interactions on each cluster and then sum the total.
- Schrödinger Suit (version 2020 or newer; comercial or academic version).
- Knime version 4.3.2 or higher, a programing software via functional workflows.
- Our in-house Knime workflow to calculate interaction frequencies. Cluster-interactions-frequency.
- A set of receptor-ligand complexes from a single cluster. All complexes must be of the same receptor and ligand. They can be obtained from different docking simulations, from a molecular dynamics trajectory, or even from free energy calculations. To test this pipeline, skipping steps 1 and 2, we provide an example ready for the workflow, with 7 clusters with all interactions already calculated (csv files), and separated into different folders. Example cluster set.
To make sure that receptor and ligand are being use as input in correct order we first split the receptor-ligand complexes into two separate files using the split_structure.py script from Schrödinger scripts. This can be done to all complex within a folder writing and executing the following bash script:
split.awk
echo "Processing data ..."
for f in *.maegz; do
$SCHRODINGER/run split_structure.py ${f} -m ligand -k -many_files ${f%.maegz}.mae
done
echo "Done!!"
Execute on a terminal with:
$ ./split.awk
For more information about the script options you can type in the terminal:
$ $SCHRODINGER/run split_structure.py -h
To calculate the interactions between receptor and ligand we use the poseviewer_interactions.py script (Schrödinger suit version 2020-1). When setting inputs make sure to give first the receptor and then the ligand file, followed by "-csv" to also export the results on a csv format file. You can calculate the interactions on all the previously split files by writing the next script:
poseviewer.awk
echo "Processing data ..."
for f in *_receptor1.mae; do
$SCHRODINGER/run poseviewer_interactions.py ${f} ${f%_receptor1.mae}_ligand1.mae -csv
done
echo "Done!!"
Execute on a terminal with:
$ ./poseviewer.awk
For more information about the script options you can type in the terminal:
$ $SCHRODINGER/run poseviewer_interactions.py -h
Make sure that each cluster has their own separated directory containing interactions files (*.csv).
$ ls
cluster1 cluster2 cluster3 cluster4 cluster5 cluster6 cluster7
$ tree
.
└── System1
├── cluster1
│ ├── 174_pv_interactions.csv
│ ├── 181_pv_interactions.csv
│ ├── 183_pv_interactions.csv
│ ├── 271_pv_interactions.csv
│ ├── 272_pv_interactions.csv
│ ├── ...
├── cluster2
│ ├── 1001_pv_interactions.csv
│ ├── 107_pv_interactions.csv
│ ├── 108_pv_interactions.csv
│ ├── 109_pv_interactions.csv
│ ├── 117_pv_interactions.csv
│ ├── ...
├── cluster3
│ ├── 12_pv_interactions.csv
│ ├── 13_pv_interactions.csv
│ ├── 143_pv_interactions.csv
│ ├── 144_pv_interactions.csv
│ ├── 145_pv_interactions.csv
│ ├── ...
└── ...
To calculate the interaction frequency of each ligand-receptor complex cluster we use the Knime workflow Multiple_clusters_interactions_frequency. The user must to configure 2 nodes:
- List Files/Folders: The user must select the parent folder where separated cluster directories are located.
- Table Creator: The user need to list all ligand atoms (use only PDB format atom names) on the first column, and assign each atom to the fragment of the structure it belongs to. Visit Interaction frequency among single-cluster conformers for an example of maping ligands.
Example of ligand fragment naming on table creator node:
The results table shows ligand framents, interacting residues and the sum of frequencies of clusters. Download results file
-
Peña-Varas, Carlos, & Ramírez, David. (2021, Sep 9). Docking post-processing: Interaction frequency among multiple-cluster conformers (Version 1.0). Zenodo. http://doi.org/10.5281/zenodo.5498074