From 1fc029a09bf1b6608b7f7609f27358c3b8c1943e Mon Sep 17 00:00:00 2001
From: Stefan Bienert <stefan.bienert@unibas.ch>
Date: Wed, 10 Jan 2024 16:41:20 +0100
Subject: [PATCH] Add software parameters
---
convert_to_modelcif.py | 152 +++++++++++++++++++++++++++++++++++++----
1 file changed, 138 insertions(+), 14 deletions(-)
diff --git a/convert_to_modelcif.py b/convert_to_modelcif.py
index bb4557e..974d594 100755
--- a/convert_to_modelcif.py
+++ b/convert_to_modelcif.py
@@ -326,21 +326,23 @@ def _get_modelcif_protocol(
)
# loop over software group and assemble software group from that
sw_grp = modelcif.SoftwareGroup()
- for pss in js_step["software_group"]: # protocol step software
- if sw_dict[pss] is not None:
+ for (
+ pss,
+ psp,
+ ) in zip( # protocol step software & protocol step parameters
+ js_step["software_group"], js_step["parameter_group"]
+ ):
+ plst = []
+ for arg, val in psp.items():
+ plst.append(modelcif.SoftwareParameter(arg, val))
+ if len(plst) > 0:
+ # add software with individual parameters
+ sw_grp.append(
+ modelcif.SoftwareWithParameters(sw_dict[pss], plst)
+ )
+ else:
# add software w/o individual parameters
sw_grp.append(sw_dict[pss])
- # add software with individual parameters
- # Commented out code does not need spelling check, so disable
- # it in Pylint
- # pylint: disable=wrong-spelling-in-comment
- # sw_grp.append(
- # modelcif.SoftwareWithParameters(
- # sw_dict[pss],
- # [modelcif.SoftwareParameter(<name>, <value>)],
- # )
- # )
- # pylint: enable=wrong-spelling-in-comment
# ToDo: make sure AlphaPulldown is first in the SoftwareGroup() list,
# AlphaFold second; that influences citation order in the ModelCIF
# file.
@@ -622,6 +624,117 @@ def _file_exists_or_exit(path, msg):
sys.exit()
+def _cast_param(val):
+ """Cast a string input val to its actual data type."""
+ try:
+ return int(val)
+ except ValueError:
+ pass
+ try:
+ return float(val)
+ except ValueError:
+ pass
+ if val == "True":
+ return True
+ if val == "False":
+ return False
+
+ return val
+
+
+def _get_software_with_parameters(sw_dict, other_dict):
+ """Get software with versions and parameters."""
+ # ToDo: deal with `use_mmseqs=True`
+ # ToDo: should all args go to AlphaPulldown?
+ known_args = {
+ "db_preset": ["AlphaFold"],
+ "max_template_date": ["AlphaFold"],
+ "model_preset": ["AlphaFold"],
+ "num_multimer_predictions_per_model": ["AlphaFold"],
+ "plddt_threshold": ["AlphaPulldown"],
+ "hb_allowance": ["AlphaPulldown"],
+ "threshold_clashes": ["AlphaPulldown"],
+ "job_index": ["AlphaPulldown"],
+ "use_mmseqs2": ["AlphaPulldown"],
+ "skip_existing": ["AlphaPulldown"],
+ "save_msa_files": ["AlphaPulldown"],
+ "num_predictions_per_model": ["AlphaPulldown"],
+ "benchmark": ["AlphaFold", "AlphaPulldown"],
+ "use_precomputed_msas": ["AlphaFold", "AlphaPulldown"],
+ "models_to_relax": ["AlphaFold", "AlphaPulldown"],
+ }
+ trans_args = {
+ "num_multimer_predictions_per_model": "num_predictions_per_model"
+ }
+ ignored_args = [
+ "?",
+ "alsologtostderr",
+ "bfd_database_path",
+ "data_dir",
+ "delta_threshold",
+ "description_file",
+ "hbm_oom_exit",
+ "hhblits_binary_path",
+ "hhsearch_binary_path",
+ "hmmbuild_binary_path",
+ "hmmsearch_binary_path",
+ "jackhmmer_binary_path",
+ "kalign_binary_path",
+ "log_dir",
+ "logger_levels",
+ "logtostderr",
+ "mgnify_database_path",
+ "obsolete_pdbs_path",
+ "only_check_args",
+ "op_conversion_fallback_to_while_loop",
+ "output_dir",
+ "path_to_fasta",
+ "path_to_mmt",
+ "pdb",
+ "pdb70_database_path",
+ "pdb_post_mortem",
+ "pdb_seqres_database_path",
+ "run_with_pdb",
+ "run_with_profiling",
+ "runtime_oom_exit",
+ "showprefixforinfo",
+ "small_bfd_database_path",
+ "stderrthreshold",
+ "template_mmcif_dir",
+ "tt_check_filter",
+ "tt_single_core_summaries",
+ "uniprot_database_path",
+ "uniref30_database_path",
+ "uniref90_database_path",
+ "use_small_bfd",
+ "v",
+ "verbosity",
+ "xml_output_file",
+ ]
+ re_args = re.compile(
+ r"(?:fasta_path|multimeric_chains|multimeric_templates|protein)_\d+"
+ )
+ swwp = sw_dict # Software With Parameters
+ for key, val in other_dict.items():
+ if key in known_args:
+ for tool in known_args[key]:
+ if "params" not in swwp[tool]:
+ swwp[tool]["params"] = {}
+ swwp[tool]["params"][f"--{key}"] = _cast_param(val)
+ if key in trans_args:
+ key = trans_args[key]
+ for tool in known_args[key]:
+ if "params" not in swwp[tool]:
+ swwp[tool]["params"] = {}
+ swwp[tool]["params"][f"--{key}"] = _cast_param(val)
+ else:
+ if key not in ignored_args and re.match(re_args, key) is None:
+ logging.info(f"Found unknown key in 'other': {key}")
+ sys.exit()
+
+ return swwp
+
+
def _get_feature_metadata(
modelcif_json: dict,
cmplx_name: str,
@@ -642,8 +755,10 @@ def _get_feature_metadata(
# ToDo: make sure that its always ASCII
with open(feature_json, "r", encoding="ascii") as jfh:
jdata = json.load(jfh)
- modelcif_json["__meta__"][mnmr]["software"] = jdata["software"]
modelcif_json["__meta__"][mnmr]["databases"] = jdata["databases"]
+ modelcif_json["__meta__"][mnmr][
+ "software"
+ ] = _get_software_with_parameters(jdata["software"], jdata["other"])
return cmplx_name
@@ -944,12 +1059,17 @@ def _get_protocol_steps(modelcif_json):
"input_data_group": ["target_sequences", "reference_dbs"],
"output_data_group": "monomer_pickle_files",
"software_group": [],
+ "parameter_group": [],
}
for sftwr in modelcif_json["__meta__"].values():
sftwr = sftwr["software"]
for tool in sftwr:
if tool not in step["software_group"]:
step["software_group"].append(tool)
+ if "params" in sftwr[tool]:
+ step["parameter_group"].append(sftwr[tool]["params"])
+ else:
+ step["parameter_group"].append({})
protocol.append(step)
# modelling step
@@ -963,6 +1083,10 @@ def _get_protocol_steps(modelcif_json):
"input_data_group": ["target_sequences", "STEPTYPE$coevolution MSA"],
"output_data_group": "model",
"software_group": ["AlphaPulldown", "AlphaFold"],
+ "parameter_group": [
+ sftwr["AlphaPulldown"]["params"],
+ sftwr["AlphaFold"]["params"],
+ ],
}
protocol.append(step)
--
GitLab