Skip to content
Snippets Groups Projects
Unverified Commit 6c3fccbd authored by Xavier Robin's avatar Xavier Robin
Browse files

feat: SCHWED-5954 add unassigned ligands to action

parent 5295fcb7
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
actions/ost-compare-ligand-structures
+ 75
− 40
View file @ 6c3fccbd
...@@ -193,6 +193,15 @@ def _ParseArgs(): ...@@ -193,6 +193,15 @@ def _ParseArgs():
action="store_true", action="store_true",
help=("Use RMSD for ligand assignment.")) help=("Use RMSD for ligand assignment."))
parser.add_argument(
"-u",
"--unassigned",
dest="unassigned",
default=False,
action="store_true",
help=("Report unassigned ligands in the output together with assigned "
"ligands."))
parser.add_argument( parser.add_argument(
"--lddt-pli", "--lddt-pli",
dest="lddt_pli", dest="lddt_pli",
...@@ -354,6 +363,7 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -354,6 +363,7 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
substructure_match=args.substructure_match, substructure_match=args.substructure_match,
global_chain_mapping=args.global_chain_mapping, global_chain_mapping=args.global_chain_mapping,
rmsd_assignment=args.rmsd_assignment, rmsd_assignment=args.rmsd_assignment,
unassigned=args.unassigned,
radius=args.radius, radius=args.radius,
lddt_pli_radius=args.lddt_pli_radius, lddt_pli_radius=args.lddt_pli_radius,
lddt_lp_radius=args.lddt_lp_radius, lddt_lp_radius=args.lddt_lp_radius,
...@@ -367,8 +377,8 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -367,8 +377,8 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
# Replace model ligand by path # Replace model ligand by path
if len(model_ligands) == len(scorer.model_ligands): if len(model_ligands) == len(scorer.model_ligands):
# Map ligand => path # Map ligand => path
model_ligands_map = {k: v for k, v in zip(scorer.model_ligands, model_ligands_map = {k.hash_code: v for k, v in zip(
args.model_ligands)} scorer.model_ligands, args.model_ligands)}
out["model_ligands"] = args.model_ligands out["model_ligands"] = args.model_ligands
elif len(model_ligands) < len(scorer.model_ligands): elif len(model_ligands) < len(scorer.model_ligands):
# Multi-ligand SDF files were given # Multi-ligand SDF files were given
...@@ -384,7 +394,7 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -384,7 +394,7 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
"(%d) than given (%d)" % ( "(%d) than given (%d)" % (
len(scorer.model_ligands), len(model_ligands))) len(scorer.model_ligands), len(model_ligands)))
else: else:
model_ligands_map = {l: _QualifiedResidueNotation(l) model_ligands_map = {l.hash_code: _QualifiedResidueNotation(l)
for l in scorer.model_ligands} for l in scorer.model_ligands}
out["model_ligands"] = list(model_ligands_map.values()) out["model_ligands"] = list(model_ligands_map.values())
...@@ -392,8 +402,8 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -392,8 +402,8 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
# Replace reference ligand by path # Replace reference ligand by path
if len(reference_ligands) == len(scorer.target_ligands): if len(reference_ligands) == len(scorer.target_ligands):
# Map ligand => path # Map ligand => path
reference_ligands_map = {k: v for k, v in zip(scorer.target_ligands, reference_ligands_map = {k.hash_code: v for k, v in zip(
args.reference_ligands)} scorer.target_ligands, args.reference_ligands)}
out["reference_ligands"] = args.reference_ligands out["reference_ligands"] = args.reference_ligands
elif len(reference_ligands) < len(scorer.target_ligands): elif len(reference_ligands) < len(scorer.target_ligands):
# Multi-ligand SDF files were given # Multi-ligand SDF files were given
...@@ -410,53 +420,78 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -410,53 +420,78 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
len(scorer.target_ligands), len(reference_ligands))) len(scorer.target_ligands), len(reference_ligands)))
else: else:
reference_ligands_map = {l: _QualifiedResidueNotation(l) reference_ligands_map = {l.hash_code: _QualifiedResidueNotation(l)
for l in scorer.target_ligands} for l in scorer.target_ligands}
out["reference_ligands"] = list(reference_ligands_map.values()) out["reference_ligands"] = list(reference_ligands_map.values())
if not (args.lddt_pli or args.rmsd):
ost.LogWarning("No score selected, output will be empty.")
else:
out["unassigned_model_ligands"] = {}
for chain, unassigned_residues in scorer.unassigned_model_ligands.items():
for resnum, unassigned in unassigned_residues.items():
mdl_lig = scorer.model.FindResidue(chain, resnum)
out["unassigned_model_ligands"][model_ligands_map[
mdl_lig.hash_code]] = unassigned
out["unassigned_reference_ligands"] = {}
for chain, unassigned_residues in scorer.unassigned_target_ligands.items():
for resnum, unassigned in unassigned_residues.items():
trg_lig = scorer.target.FindResidue(chain, resnum)
out["unassigned_reference_ligands"][reference_ligands_map[
trg_lig.hash_code]] = unassigned
if args.lddt_pli: if args.lddt_pli:
out["lddt_pli"] = {} out["lddt_pli"] = {}
for chain, lddt_pli_results in scorer.lddt_pli_details.items(): for chain, lddt_pli_results in scorer.lddt_pli_details.items():
for _, lddt_pli in lddt_pli_results.items(): for resnum, lddt_pli in lddt_pli_results.items():
model_key = model_ligands_map[lddt_pli["model_ligand"]] if args.unassigned and lddt_pli["unassigned"]:
lddt_pli["reference_ligand"] = reference_ligands_map[ mdl_lig = scorer.model.FindResidue(chain, resnum)
lddt_pli.pop("target_ligand")] model_key = model_ligands_map[mdl_lig.hash_code]
lddt_pli["model_ligand"] = model_key else:
transform_data = lddt_pli["transform"].data model_key = model_ligands_map[lddt_pli["model_ligand"].hash_code]
lddt_pli["transform"] = [transform_data[i:i + 4] lddt_pli["reference_ligand"] = reference_ligands_map[
for i in range(0, len(transform_data), lddt_pli.pop("target_ligand").hash_code]
4)] lddt_pli["model_ligand"] = model_key
lddt_pli["bs_ref_res"] = [_QualifiedResidueNotation(r) for r in transform_data = lddt_pli["transform"].data
lddt_pli["bs_ref_res"]] lddt_pli["transform"] = [transform_data[i:i + 4]
lddt_pli["bs_ref_res_mapped"] = [_QualifiedResidueNotation(r) for r in for i in range(0, len(transform_data),
lddt_pli["bs_ref_res_mapped"]] 4)]
lddt_pli["bs_mdl_res_mapped"] = [_QualifiedResidueNotation(r) for r in lddt_pli["bs_ref_res"] = [_QualifiedResidueNotation(r) for r in
lddt_pli["bs_mdl_res_mapped"]] lddt_pli["bs_ref_res"]]
lddt_pli["inconsistent_residues"] = ["%s-%s" %( lddt_pli["bs_ref_res_mapped"] = [_QualifiedResidueNotation(r) for r in
_QualifiedResidueNotation(x), _QualifiedResidueNotation(y)) for x,y in lddt_pli[ lddt_pli["bs_ref_res_mapped"]]
"inconsistent_residues"]] lddt_pli["bs_mdl_res_mapped"] = [_QualifiedResidueNotation(r) for r in
lddt_pli["bs_mdl_res_mapped"]]
lddt_pli["inconsistent_residues"] = ["%s-%s" %(
_QualifiedResidueNotation(x), _QualifiedResidueNotation(y)) for x,y in lddt_pli[
"inconsistent_residues"]]
out["lddt_pli"][model_key] = lddt_pli out["lddt_pli"][model_key] = lddt_pli
if args.rmsd: if args.rmsd:
out["rmsd"] = {} out["rmsd"] = {}
for chain, rmsd_results in scorer.rmsd_details.items(): for chain, rmsd_results in scorer.rmsd_details.items():
for _, rmsd in rmsd_results.items(): for _, rmsd in rmsd_results.items():
model_key = model_ligands_map[rmsd["model_ligand"]] if args.unassigned and rmsd["unassigned"]:
rmsd["reference_ligand"] = reference_ligands_map[ mdl_lig = scorer.model.FindResidue(chain, resnum)
rmsd.pop("target_ligand")] model_key = model_ligands_map[mdl_lig.hash_code]
rmsd["model_ligand"] = model_key else:
transform_data = rmsd["transform"].data model_key = model_ligands_map[rmsd["model_ligand"].hash_code]
rmsd["transform"] = [transform_data[i:i + 4] rmsd["reference_ligand"] = reference_ligands_map[
for i in range(0, len(transform_data), 4)] rmsd.pop("target_ligand").hash_code]
rmsd["bs_ref_res"] = [_QualifiedResidueNotation(r) for r in rmsd["model_ligand"] = model_key
rmsd["bs_ref_res"]] transform_data = rmsd["transform"].data
rmsd["bs_ref_res_mapped"] = [_QualifiedResidueNotation(r) for r in rmsd["transform"] = [transform_data[i:i + 4]
rmsd["bs_ref_res_mapped"]] for i in range(0, len(transform_data), 4)]
rmsd["bs_mdl_res_mapped"] = [_QualifiedResidueNotation(r) for r in rmsd["bs_ref_res"] = [_QualifiedResidueNotation(r) for r in
rmsd["bs_mdl_res_mapped"]] rmsd["bs_ref_res"]]
rmsd["inconsistent_residues"] = ["%s-%s" %( rmsd["bs_ref_res_mapped"] = [_QualifiedResidueNotation(r) for r in
_QualifiedResidueNotation(x), _QualifiedResidueNotation(y)) for x,y in rmsd[ rmsd["bs_ref_res_mapped"]]
"inconsistent_residues"]] rmsd["bs_mdl_res_mapped"] = [_QualifiedResidueNotation(r) for r in
rmsd["bs_mdl_res_mapped"]]
rmsd["inconsistent_residues"] = ["%s-%s" %(
_QualifiedResidueNotation(x), _QualifiedResidueNotation(y)) for x,y in rmsd[
"inconsistent_residues"]]
out["rmsd"][model_key] = rmsd out["rmsd"][model_key] = rmsd
return out return out
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment