Original receptor shifts significantly after calling reduce2

[Olabisi-Aishat-Bello]

Hi,

I'm trying to run hydrogen optimization on these PDB files '8hg7' and '8hez'. However, after calling reduce2 on a PDB file that has my selected chain, the chain is now significantly shifted away from its initial coordinates such that even the ligand it was crystallized with is no longer in the same coordinates.

This is the example code that reproduces the error. I have also included the output image visualization file before and after optimization.

pdb_token = '8hez'
pdb_file = '.../Vina/sglt2/8hez.pdb'
atoms_from_pdb = parsePDB(pdb_file)
receptor_selection = "chain A and not water and not hetero"
receptor_atoms = atoms_from_pdb.select(receptor_selection)
prody_receptorPDB = f"{pdb_token}_receptor_atoms.pdb"
writePDB(prody_receptorPDB, receptor_atoms)

reduce_inputPDB = f"{pdb_token}_receptor.pdb"

subprocess.run(
f'cat <(grep "CRYST1" {pdb_file}) {prody_receptorPDB} > {reduce_inputPDB}',
shell=True,
executable="/bin/bash"
)

reduce_opts = "approach=add add_flip_movers=True"
overwrite = "overwrite=True"

command = f'export MMTBX_CCP4_MONOMER_LIB="{geostd_path}"; python {reduce2} {reduce_inputPDB} {reduce_opts} {overwrite}'
subprocess.run(command, shell=True, executable="/bin/bash")

prepare_inPDB = f"{pdb_token}_receptorFH.pdb"

atoms_from_pdb = parsePDB(pdb_file)
ligand_selection = "chain A and resname LE6"
ligand_atoms = atoms_from_pdb.select(ligand_selection)
center_x, center_y, center_z = calcCenter(ligand_atoms)

size_x = 20.0
size_y = 20.0
size_z = 20.0

prody_ligandPDB = "LIG.pdb"
writePDB(prody_ligandPDB, ligand_atoms)

[olegsobolev]

Hi,

I don't know what you do in parsePDB() and writePDB() but assuming nothing too exciting:

Most likely what you see is inconsistency in CRYST1 records between the files. The original 8hez.pdb when taken from PDB has
CRYST1 1.000 1.000 1.000 90.00 90.00 90.00 P 1

It is nonsensical but typical for CryoEM models. Likely reduce2 puts the model in artificial box and outputs CRYST1 with some meaningful values. Check the files. I'd put original model into the box if needed explicitly and work with it.
Something like:

import iotbx.pdb
import mmtbx.model
from libtbx import easy_run

pdb_token = '8hez'
pdb_file = '8hez.pdb'

# Load model
original_model = mmtbx.model.manager(model_input = iotbx.pdb.input(pdb_file))
# Put box around it and write it out if needed:
original_model.add_crystal_symmetry_if_necessary()
with open("dummy_symmetry.pdb", 'w') as f:
  f.write(original_model.model_as_pdb())

# Dump receptor part
receptor_selection = original_model.selection("chain A and not water and not hetero")
receptor_model = original_model.select(receptor_selection)
reduce_inputPDB = f"{pdb_token}_receptor.pdb"
with open(reduce_inputPDB, 'w') as f:
  f.write(receptor_model.model_as_pdb())

# run reduce2 on receptor
reduce_opts = "approach=add add_flip_movers=True"
overwrite = "overwrite=True"
reduce2 = "mmtbx.reduce2"
command = f'{reduce2} {reduce_inputPDB} {reduce_opts} {overwrite}'
easy_run.call(command)

# Dump ligand from original model
lig_selection = original_model.selection("chain A and resname LE6")
lig_model = original_model.select(lig_selection)
prody_ligandPDB = "LIG.pdb"
with open(prody_ligandPDB, 'w') as f:
  f.write(lig_model.model_as_pdb())