DesignConstraints.to_dict

DesignConstraints.to_dict()[source]

Export the constraint set as a plain, JSON-serializable dictionary.

The returned dictionary holds one key per constraint field plus parent, with copies of the stored containers, so mutating it never changes the object. It is the input of DesignConstraints.from_dict(), which reconstructs an equal object.

Returns:: dict_constraints – One key per field of ut.LIST_DESIGN_CONSTRAINTS plus parent; unset fields are None.
Return type:: dict

See also

DesignConstraints.from_dict(): the inverse.

Added in version 1.2.0.

Examples

:meth:DesignConstraints.to_dict exports the constraint set as a plain, JSON-serializable dictionary with one key per field plus parent, so a design campaign can be written to disk, logged next to its results, or handed to another process. It takes no parameters, and :meth:DesignConstraints.from_dict reconstructs an equal object from its output.

import json

import pandas as pd
import aaanalysis as aa
aa.options["verbose"] = False

df_seq = aa.load_dataset(name="DOM_GSEC", n=5)
seq = df_seq["sequence"].iloc[0]
tmd_start, tmd_stop = int(df_seq["tmd_start"].iloc[0]), int(df_seq["tmd_stop"].iloc[0])

dc = aa.DesignConstraints(immutable_positions=[tmd_start, tmd_stop],
                          permitted_substitutions=["A", "L", "V", "I"],
                          forbidden_substitutions={tmd_start + 6: ["V"]},
                          n_mut_max=2,
                          min_identity=0.95,
                          forbidden_motifs=["WW"],
                          parent=seq)
dict_constraints = dc.to_dict()
df_constraints = pd.DataFrame({"field": list(dict_constraints),
                               "value": [str(v)[:60] for v in dict_constraints.values()]})
aa.display_df(df_constraints, n_rows=10, show_shape=True)

DataFrame shape: (10, 2)

	field	value
1	immutable_positions	[37, 59]
2	mutable_positions	None
3	permitted_substitutions	['A', 'L', 'V', 'I']
4	forbidden_substitutions	{43: ['V']}
5	n_mut_max	2
6	min_identity	0.95
7	max_identity	None
8	forbidden_motifs	['WW']
9	required_motifs	None
10	parent	MQKVTLGLLVFLAGF...GVLCAMGIIIVMSAK

Unset fields are None, and the containers are copies, so editing the exported dictionary never changes the object it came from. The dictionary survives a JSON round trip: from_dict converts the integer position keys that JSON turned into digit strings back.

text = json.dumps(dict_constraints)
dict_constraints["n_mut_max"] = 99           # editing the export leaves the object untouched
dc_json = aa.DesignConstraints.from_dict(dict_constraints=json.loads(text))

df_roundtrip = pd.DataFrame([dict(step="object", n_mut_max=dc.n_mut_max, json_chars=len(text)),
                             dict(step="edited export", n_mut_max=dict_constraints["n_mut_max"],
                                  json_chars=len(text)),
                             dict(step="from JSON", n_mut_max=dc_json.n_mut_max, json_chars=len(text))])
aa.display_df(df_roundtrip, n_rows=10, show_shape=True)
print("round trip equal:", dc_json == dc)

DataFrame shape: (3, 3)

	step	n_mut_max	json_chars
1	object	2	365
2	edited export	99	365
3	from JSON	2	365

round trip equal: True