SLAM

SPACIO-LINEAR SCREENING OF LIGAND-DOCKING CAVITIES IN PROTEIN STRUCTURES: SLAM approach

The SLAM approach assesses the similarity between a probe ligand-binding cavity and a database of target protein cavities. SLAM takes as input a PDB file containing a ligand–protein complex, extracts the ligand-binding cavity (neighborhood), and aligns it against a library of pre-generated cavities from target proteins in the PDB database. The algorithm represents the 3D structures of both the query ligand-binding cavity and the target protein cavities as sets of vectors encoding the physicochemical properties of their atoms. By comparing these sets of linear vectors and analyzing their correlations, SLAM enables fast and efficient local comparison of 3D protein structures.

SLAM:

Step1:

Generate probe-ligand neighborhood using getLigandHetChain program

Step2:

Generate database of target proteins

• Option1: LigandDB: database of all ligand-binding cavities from PDB structures
  1. Using getLigandHetChain program on a list of target proteins getLigandHetChain program to generate all target protein ligand neighborhoods
  2. Combine all target protein ligand neighborhoods into one sdf file LigandDB.sdf

        cat *Dist#7_Ch* > LigandDB.sdf

• Option2: SurfaceXDB: database of solvent accesible surface neighborhoods in all PDB proteins
  1. Using freesasarun.py execute FreeSASA on all PDB structures freesasarun.py
  2. Using add_chain_x.py generate AX PDB files add_chain_x.py
  3. Using SurfaceXPDB generate a database of surfaces of all target proteins SurfaceXPDB
  4. Combine all target protein surface neighborhoods into one sdf file SurfaceXDB.sdf

        cat *Dist#7_Ch* > SurfaceXDB.sdf

Step3: SLAM screening of probe-ligand vs. target neighborhood DB (LigandDB or SurfaceDB)

The step takes as input probe-ligand neighborhood and LigandDB (or SurfaceXDB) database utilizes AlgLocPhyChemStruct to find matches between probe-ligand neighborhood and LigandDB (or SurfaceXDB) target proteins.

Step4: Filtering SLAM results with NAtom*Corr5>threshold

This step uses align_filter Python script to filter potentially good matches between probe-ligand neighborhood and target proteins.

Step5: Superimposition of Target-protein neighborhood on Probe-protein

This step uses lig_choice_superimpose Python script to prepare a PDB file representing a molecular complex consisting of probe protein, ligand and target protein.

Step 6: Vina calculations

This step uses lig_lig_vina script to run Vina calculations of free energy and save them into output table.

Step 7: FE-score and filtering