Skip to content
Snippets Groups Projects
Commit f5fcccb5 authored by Studer Gabriel's avatar Studer Gabriel
Browse files

compare-ligand-structures refactoring

parent 014ebc2e
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
+ 150
− 83
View file @ f5fcccb5
...@@ -47,23 +47,29 @@ options, this is a dictionary with three keys: ...@@ -47,23 +47,29 @@ options, this is a dictionary with three keys:
content of the JSON output will be \"status\" set to FAILURE and an content of the JSON output will be \"status\" set to FAILURE and an
additional key: "traceback". additional key: "traceback".
Each score is opt-in and the respective results are available in two keys: Each score is opt-in and the respective results are available in three keys:
* "model_ligands": Model ligand centric scoring based on model/reference * "assigned_scores": A list with data for each pair of assigned ligands.
ligand assignment. A score including meta data is reported for each assigned Data is yet another dict containing score specific information for that
model ligand given the assigned target ligand. Unassigned model ligands are ligand pair. The following keys are there in any case:
reported with a null score and a reason why no assignment has been performed.
* "model_ligand": The model ligand
* "full": The full all vs. all scoring results. Yet another dictionary with * "reference_ligand": The target ligand to which model ligand is assigned to
keys: * "score": The score
* "coverage": Fraction of model ligand atoms which are covered by target
* "assignment": List of pairs in form (ref_lig_idx, mdl_lig_idx) specifying ligand. Will only deviate from 1.0 if --substructure-match is enabled.
the ligands in "reference_ligands"/"model_ligands".
* "scores": A dictionary with key in form (ref_lig_idx, mdl_lig_idx) and * "model_ligand_unassigned_reason": Dictionary with unassigned model ligands
value yet another dict with score information for each possible as key and an educated guess why this happened.
reference/model ligand pair. The respective score is None if no score could
be computed. This can simply be a mismatch between the two ligands. This or * "reference_ligand_unassigned_reason": Dictionary with unassigned target ligands
other reasons are reported. as key and an educated guess why this happened.
If --full-results is enabled, another element with key "full_results" is added.
This is a list of data items for each pair of model/reference ligands. The data
items follow the same structure as in "assigned_scores". If no score for a
specific pair of ligands could be computed, "score" and "coverage" are set to
null and a key "reason" is added giving an educated guess why this happened.
""" """
import argparse import argparse
...@@ -212,6 +218,14 @@ def _ParseArgs(): ...@@ -212,6 +218,14 @@ def _ParseArgs():
default=3, default=3,
help="Set verbosity level. Defaults to 3 (INFO).") help="Set verbosity level. Defaults to 3 (INFO).")
parser.add_argument(
"--full-results",
dest="full_results",
default=False,
action="store_true",
help=("Outputs scoring results for all model/reference ligand pairs "
"and store as key \"full_results\""))
# arguments relevant for lddt-pli # arguments relevant for lddt-pli
parser.add_argument( parser.add_argument(
...@@ -219,7 +233,7 @@ def _ParseArgs(): ...@@ -219,7 +233,7 @@ def _ParseArgs():
dest="lddt_pli", dest="lddt_pli",
default=False, default=False,
action="store_true", action="store_true",
help=("Compute lDDT-PLI score and store as key \"lddt-pli\".")) help=("Compute lDDT-PLI scores and store as key \"lddt_pli\"."))
parser.add_argument( parser.add_argument(
"--lddt-pli-radius", "--lddt-pli-radius",
...@@ -241,7 +255,7 @@ def _ParseArgs(): ...@@ -241,7 +255,7 @@ def _ParseArgs():
dest="rmsd", dest="rmsd",
default=False, default=False,
action="store_true", action="store_true",
help=("Compute RMSD score and store as key \"rmsd\".")) help=("Compute RMSD scores and store as key \"rmsd\"."))
parser.add_argument( parser.add_argument(
"--radius", "--radius",
...@@ -514,48 +528,69 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -514,48 +528,69 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
if args.lddt_pli: if args.lddt_pli:
out["lddt_pli"] = dict() out["lddt_pli"] = dict()
out["lddt_pli"]["model_ligands"] = dict() out["lddt_pli"]["assigned_scores"] = list()
out["lddt_pli"]["full"] = dict()
for lig_pair in lddtpli_scorer.assignment: for lig_pair in lddtpli_scorer.assignment:
score = float(lddtpli_scorer.score_matrix[lig_pair[0], lig_pair[1]]) score = float(lddtpli_scorer.score_matrix[lig_pair[0], lig_pair[1]])
coverage = float(lddtpli_scorer.coverage_matrix[lig_pair[0], lig_pair[1]]) coverage = float(lddtpli_scorer.coverage_matrix[lig_pair[0], lig_pair[1]])
aux_data = lddtpli_scorer.aux_matrix[lig_pair[0], lig_pair[1]] aux_data = lddtpli_scorer.aux_matrix[lig_pair[0], lig_pair[1]]
target_key = out["reference_ligands"][lig_pair[0]] target_key = out["reference_ligands"][lig_pair[0]]
model_key = out["model_ligands"][lig_pair[1]] model_key = out["model_ligands"][lig_pair[1]]
out["lddt_pli"]["model_ligands"][model_key] = {"lddt_pli": score, out["lddt_pli"]["assigned_scores"].append({"score": score,
"coverage": coverage, "coverage": coverage,
"lddt_pli_n_contacts": aux_data["lddt_pli_n_contacts"], "lddt_pli_n_contacts": aux_data["lddt_pli_n_contacts"],
"model_ligand": model_key, "model_ligand": model_key,
"reference_ligand": target_key, "reference_ligand": target_key,
"bs_ref_res": [_QualifiedResidueNotation(r) for r in "bs_ref_res": [_QualifiedResidueNotation(r) for r in
aux_data["bs_ref_res"]], aux_data["bs_ref_res"]],
"bs_mdl_res": [_QualifiedResidueNotation(r) for r in "bs_mdl_res": [_QualifiedResidueNotation(r) for r in
aux_data["bs_mdl_res"]]} aux_data["bs_mdl_res"]]})
out["lddt_pli"]["model_ligand_unassigned_reason"] = dict()
for i in lddtpli_scorer.unassigned_model_ligands: for i in lddtpli_scorer.unassigned_model_ligands:
model_key = out["model_ligands"][i] key = out["model_ligands"][i]
reason = lddtpli_scorer.guess_model_ligand_unassigned_reason(i) reason = lddtpli_scorer.guess_model_ligand_unassigned_reason(i)
out["lddt_pli"]["model_ligands"][model_key] = {"lddt_pli": None, out["lddt_pli"]["model_ligand_unassigned_reason"][key] = reason
"unassigned_reason": reason}
out["lddt_pli"]["reference_ligand_unassigned_reason"] = dict()
out["lddt_pli"]["full"]["assignment"] = lddtpli_scorer.assignment for i in lddtpli_scorer.unassigned_target_ligands:
out["lddt_pli"]["full"]["scores"] = dict() key = out["reference_ligands"][i]
reason = lddtpli_scorer.guess_target_ligand_unassigned_reason(i)
shape = lddtpli_scorer.score_matrix.shape out["lddt_pli"]["reference_ligand_unassigned_reason"][key] = reason
for ref_lig_idx in range(shape[0]):
for mdl_lig_idx in range(shape[1]): if args.full_results:
score = float(lddtpli_scorer.score_matrix[(ref_lig_idx, mdl_lig_idx)]) out["lddt_pli"]["full_results"] = list()
state = int(lddtpli_scorer.state_matrix[(ref_lig_idx, mdl_lig_idx)]) shape = lddtpli_scorer.score_matrix.shape
desc = lddtpli_scorer.state_decoding[state] for ref_lig_idx in range(shape[0]):
pair_key = [ref_lig_idx, mdl_lig_idx] for mdl_lig_idx in range(shape[1]):
out["lddt_pli"]["full"]["scores"][pair_key] = {"score": score, state = int(lddtpli_scorer.state_matrix[(ref_lig_idx, mdl_lig_idx)])
"state": desc} target_key = out["reference_ligands"][lig_pair[0]]
model_key = out["model_ligands"][lig_pair[1]]
if state == 0:
score = float(lddtpli_scorer.score_matrix[(ref_lig_idx, mdl_lig_idx)])
coverage = float(lddtpli_scorer.coverage_matrix[(ref_lig_idx, mdl_lig_idx)])
aux_data = lddtpli_scorer.aux_matrix[(ref_lig_idx, mdl_lig_idx)]
out["lddt_pli"]["full_results"].append({"score": score,
"coverage": coverage,
"lddt_pli_n_contacts": aux_data["lddt_pli_n_contacts"],
"model_ligand": model_key,
"reference_ligand": target_key,
"bs_ref_res": [_QualifiedResidueNotation(r) for r in
aux_data["bs_ref_res"]],
"bs_mdl_res": [_QualifiedResidueNotation(r) for r in
aux_data["bs_mdl_res"]]})
else:
reason = lddtpli_scorer.state_decoding[state]
out["lddt_pli"]["full_results"].append({"score": None,
"coverage": None,
"model_ligand": model_key,
"reference_ligand": target_key,
"reason": reason})
if args.rmsd: if args.rmsd:
out["rmsd"] = dict() out["rmsd"] = dict()
out["rmsd"]["model_ligands"] = dict() out["rmsd"]["assigned_scores"] = list()
out["rmsd"]["full"] = dict()
for lig_pair in scrmsd_scorer.assignment: for lig_pair in scrmsd_scorer.assignment:
score = float(scrmsd_scorer.score_matrix[lig_pair[0], lig_pair[1]]) score = float(scrmsd_scorer.score_matrix[lig_pair[0], lig_pair[1]])
coverage = float(scrmsd_scorer.coverage_matrix[lig_pair[0], lig_pair[1]]) coverage = float(scrmsd_scorer.coverage_matrix[lig_pair[0], lig_pair[1]])
...@@ -563,42 +598,74 @@ def _Process(model, model_ligands, reference, reference_ligands, args): ...@@ -563,42 +598,74 @@ def _Process(model, model_ligands, reference, reference_ligands, args):
target_key = out["reference_ligands"][lig_pair[0]] target_key = out["reference_ligands"][lig_pair[0]]
model_key = out["model_ligands"][lig_pair[1]] model_key = out["model_ligands"][lig_pair[1]]
transform_data = aux_data["transform"].data transform_data = aux_data["transform"].data
out["rmsd"]["model_ligands"][model_key] = {"rmsd": score, out["rmsd"]["assigned_scores"].append({"score": score,
"coverage": coverage, "coverage": coverage,
"lddt_lp": aux_data["lddt_lp"], "lddt_lp": aux_data["lddt_lp"],
"bb_rmsd": aux_data["bb_rmsd"], "bb_rmsd": aux_data["bb_rmsd"],
"model_ligand": model_key, "model_ligand": model_key,
"reference_ligand": target_key, "reference_ligand": target_key,
"chain_mapping": aux_data["chain_mapping"], "chain_mapping": aux_data["chain_mapping"],
"bs_ref_res": [_QualifiedResidueNotation(r) for r in "bs_ref_res": [_QualifiedResidueNotation(r) for r in
aux_data["bs_ref_res"]], aux_data["bs_ref_res"]],
"bs_ref_res_mapped": [_QualifiedResidueNotation(r) for r in "bs_ref_res_mapped": [_QualifiedResidueNotation(r) for r in
aux_data["bs_ref_res_mapped"]], aux_data["bs_ref_res_mapped"]],
"bs_mdl_res_mapped": [_QualifiedResidueNotation(r) for r in "bs_mdl_res_mapped": [_QualifiedResidueNotation(r) for r in
aux_data["bs_mdl_res_mapped"]], aux_data["bs_mdl_res_mapped"]],
"inconsistent_residues": [_QualifiedResidueNotation(r) for r in "inconsistent_residues": [_QualifiedResidueNotation(r) for r in
aux_data["inconsistent_residues"]], aux_data["inconsistent_residues"]],
"transform": [transform_data[i:i + 4] "transform": [transform_data[i:i + 4]
for i in range(0, len(transform_data), 4)]} for i in range(0, len(transform_data), 4)]})
out["rmsd"]["model_ligand_unassigned_reason"] = dict()
for i in scrmsd_scorer.unassigned_model_ligands: for i in scrmsd_scorer.unassigned_model_ligands:
model_key = out["model_ligands"][i] key = out["model_ligands"][i]
reason = scrmsd_scorer.guess_model_ligand_unassigned_reason(i) reason = scrmsd_scorer.guess_model_ligand_unassigned_reason(i)
out["rmsd"]["model_ligands"][model_key] = {"rmsd": None, out["rmsd"]["model_ligand_unassigned_reason"][key] = reason
"unassigned_reason": reason}
out["rmsd"]["reference_ligand_unassigned_reason"] = dict()
out["rmsd"]["full"]["assignment"] = scrmsd_scorer.assignment for i in scrmsd_scorer.unassigned_target_ligands:
out["rmsd"]["full"]["scores"] = dict() key = out["reference_ligands"][i]
reason = scrmsd_scorer.guess_target_ligand_unassigned_reason(i)
shape = scrmsd_scorer.score_matrix.shape out["rmsd"]["reference_ligand_unassigned_reason"][key] = reason
for ref_lig_idx in range(shape[0]):
for mdl_lig_idx in range(shape[1]): if args.full_results:
score = float(scrmsd_scorer.score_matrix[(ref_lig_idx, mdl_lig_idx)]) out["rmsd"]["full_results"] = list()
state = int(scrmsd_scorer.state_matrix[(ref_lig_idx, mdl_lig_idx)]) shape = scrmsd_scorer.score_matrix.shape
desc = scrmsd_scorer.state_decoding[state] for ref_lig_idx in range(shape[0]):
pair_key = [ref_lig_idx, mdl_lig_idx] for mdl_lig_idx in range(shape[1]):
out["rmsd"]["full"]["scores"][pair_key] = {"score": score, state = int(scrmsd_scorer.state_matrix[(ref_lig_idx, mdl_lig_idx)])
"state": desc} target_key = out["reference_ligands"][lig_pair[0]]
model_key = out["model_ligands"][lig_pair[1]]
if state == 0:
score = float(scrmsd_scorer.score_matrix[(ref_lig_idx, mdl_lig_idx)])
coverage = float(scrmsd_scorer.coverage_matrix[(ref_lig_idx, mdl_lig_idx)])
aux_data = scrmsd_scorer.aux_matrix[(ref_lig_idx, mdl_lig_idx)]
transform_data = aux_data["transform"].data
out["rmsd"]["full_results"].append({"score": score,
"coverage": coverage,
"lddt_lp": aux_data["lddt_lp"],
"bb_rmsd": aux_data["bb_rmsd"],
"model_ligand": model_key,
"reference_ligand": target_key,
"chain_mapping": aux_data["chain_mapping"],
"bs_ref_res": [_QualifiedResidueNotation(r) for r in
aux_data["bs_ref_res"]],
"bs_ref_res_mapped": [_QualifiedResidueNotation(r) for r in
aux_data["bs_ref_res_mapped"]],
"bs_mdl_res_mapped": [_QualifiedResidueNotation(r) for r in
aux_data["bs_mdl_res_mapped"]],
"inconsistent_residues": [_QualifiedResidueNotation(r) for r in
aux_data["inconsistent_residues"]],
"transform": [transform_data[i:i + 4]
for i in range(0, len(transform_data), 4)]})
else:
reason = scrmsd_scorer.state_decoding[state]
out["rmsd"]["full_results"].append({"score": None,
"coverage": None,
"model_ligand": model_key,
"reference_ligand": target_key,
"reason": reason})
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