BindingNet Database

The BindingNet aims at modeling high-quality binding poses for protein-ligand complexes with experimentally determined binding affinity data. BindingNet provides valuable insights into investigating protein-ligand interactions, allowing visual inspection and interpretation of structural analogs' structure-activity relationships (SARs). It can also be used for evaluating machine learning-based scoring functions and has the potential utilization for benchmarking the molecular docking methods and ligand binding free energy calculation approaches.

By comparative complex stricture modeling, it now contains 69,816 modeled high-quality protein-ligand complex structures with experimental binding affinity data from ChEMBL_v28 and template structures from PDBbind_v2019.

BindingNet is available at http://bindingnet.huanglab.org.cn/ under a CC BY-NC 4.0 license.

1. Enviroment

• build database: conda env create -f bindingnet_generate.yml

  2. Workflow

  2.0 Download PDBbind_v2019 dataset

  2.1 Extract PDB ID in PDBbind

• cd 1-extract_PDBid
• bash extractPDBid.sh -> PDBIDs_INDEX_general_PL_data.2019

  2.2 Map PDB ID to Uniprot ID

• Retrieve/ID mapping tool
• Upload the file PDBIDs_INDEX_general_PL_data.2019
• select from PDB to UniPortKB in Select options and click Submit
• Colunms: Your list...(PDB ID), Entry, ChEMBL, (BindingDB), Protein names
• Download: Tab-seperated -> 2-query_target_ChEMBLid/converted_PDBIDs_INDEX_general_PL_data.2019.tab
• change the name of column #1 to PDB ID
• run query select * where a3 !== "" in RBQL Console (VSCode extension Rainbow CSV)
  • Ctrl + Shift + P at VSCode
  • Rainbow CSV: RBQL
  • select * where a3 !== "" -> 2-query_target_ChEMBLid/converted_PDBIDs_INDEX_general_PL_data.2019.tab.tsv

    2.3 Search corresponding ligand for each receptor which has unique UniPort ID from ChEMBL

• cd 3-query_ChEMBL
• python query_chembl_v2019_x019.py
  • ChEMBL database in lab
  • ChEMBL webresource client
  • support multithreaded for high I/O

    2.4 Find compounds similar with crystal ligand within each target

• cd 4-extract_similar_compnds
• bash extract_simi_compounds.sh

  2.5 Align - Filter Serious Clash - Rescore - Filter by energy - Calculate core RMSD - Extract final pose - Add Affinity

• cd 5-pipeline_after_4-extract_simi_compnds

  2.5.1 obtain all target_pdbid_compound list

• cd 1-obtain_list
• bash obtain_target_pdbid_list.sh -> all_target_pdbid.list -> for rec_opt
• bash obtain_target_pdbid_compound_list.sh -> all_target_pdbid_compound.list

  2.5.2 receptor h opt yes in plop6.0

• cd 2-rec_opt
• bash rec_opt_qsub_anywhere.sh

  2.5.3 align - filter Serious Clash - rescore - filter by energy - calculate core RMSD - extract final pose

• cd 3-align-filter_clash-rescore-final_filter
• qsub -p -100 run_for_each_compound.sh
• task array: -t start-end
  • ${SGE_TASK_ID}: "target pdbid_compound_id"
  • less than 75000 tasks once
  • split 195686 tasks into 3 fold
  • 70,000 tasks for each script

    2.5.4 Output

• rec_h_opt.pdb
• CHEMBLxxx_xxxx_final.csv
• CHEMBLxxx_xxxx_dlig_xxx_dtotal_xxxCoreRMSD_xxx_ene.csv
• CHEMBLxxx_xxxx_dlig_xxx_dtotal_xxxCoreRMSD_xxx_final.pdb

  2.6 Deal with result: generate PLANet_all/PLANet_Uw index

• cd 5-pipeline_after_4-extract_simi_compnds/4-deal_with_result

  2.7 Requery And Obtain all activities for SAR

• cd 5-pipeline_after_4-extract_simi_compnds/5-Requery_And_Obtain_all_affinity_for_SAR

  2.8 Convert _final.pdb to compound.sdf, and Extract pocket for machine learning

• cd 5-pipeline_after_4-extract_simi_compnds/6-convert_sdf_AND_extract_pocket

  2.9 PDBbind minimized

• cd 5-pipeline_after_4-extract_simi_compnds/7-PDBbind_v2019_minimize

  2.10 Train on SGCNN model

• cd 6-deep_learning/2-FAST/

  2.11 Construct web_server

• cd 7-web_server/

  2.12 Analysis property distribution

• cd 10-analysis

  Citation

  Li, X.; Shen, C.; Zhu, H.; Yang, Y.; Wang, Q.; Yang, J.; Huang, N. A High-Quality Data Set of Protein–Ligand Binding Interactions Via Comparative Complex Structure Modeling. J. Chem. Inf. Model. 2024. https://doi.org/10.1021/acs.jcim.3c01170 .