1
|
|
|
from __future__ import annotations |
|
|
|
|
2
|
|
|
|
3
|
|
|
import logging |
|
|
|
|
4
|
|
|
from dataclasses import dataclass |
5
|
|
|
from typing import Sequence |
6
|
|
|
|
7
|
|
|
from pocketutils.core.dot_dict import NestedDotDict |
|
|
|
|
8
|
|
|
|
9
|
|
|
from mandos import logger |
|
|
|
|
10
|
|
|
from mandos.model.chembl_api import ChemblApi |
11
|
|
|
from mandos.model.chembl_support import ChemblCompound |
12
|
|
|
from mandos.model.chembl_support.chembl_utils import ChemblUtils |
13
|
|
|
from mandos.search.chembl import ChemblHit, ChemblSearch |
14
|
|
|
|
15
|
|
|
|
16
|
|
|
@dataclass(frozen=True, order=True, repr=True) |
17
|
|
|
class IndicationHit(ChemblHit): |
18
|
|
|
""" |
19
|
|
|
An indication with a MESH term. |
20
|
|
|
""" |
21
|
|
|
|
22
|
|
|
max_phase: int |
23
|
|
|
first_approval_year: str |
24
|
|
|
|
25
|
|
|
|
26
|
|
|
class IndicationSearch(ChemblSearch[IndicationHit]): |
|
|
|
|
27
|
|
|
"""""" |
28
|
|
|
|
29
|
|
|
def __init__(self, key: str, api: ChemblApi, min_phase: int): |
30
|
|
|
super().__init__(key, api) |
31
|
|
|
self.min_phase = min_phase |
32
|
|
|
|
33
|
|
|
@property |
34
|
|
|
def data_source(self) -> str: |
|
|
|
|
35
|
|
|
return "ChEMBL :: indications" |
36
|
|
|
|
37
|
|
|
def find(self, lookup: str) -> Sequence[IndicationHit]: |
38
|
|
|
""" |
39
|
|
|
|
40
|
|
|
Args: |
41
|
|
|
lookup: |
42
|
|
|
|
43
|
|
|
Returns: |
44
|
|
|
|
45
|
|
|
""" |
46
|
|
|
# 'atc_classifications': ['S01HA01', 'N01BC01', 'R02AD03', 'S02DA02'] |
47
|
|
|
# 'indication_class': 'Anesthetic (topical)' |
48
|
|
|
ch = ChemblUtils(self.api).get_compound_dot_dict(lookup) |
|
|
|
|
49
|
|
|
compound = ChemblUtils(self.api).compound_dot_dict_to_obj(ch) |
50
|
|
|
inds = self.api.drug_indication.filter(parent_molecule_chembl_id=compound.chid) |
51
|
|
|
hits = [] |
52
|
|
|
for ind in inds: |
53
|
|
|
if ind.req_as("max_phase_for_ind", int) >= self.min_phase: |
54
|
|
|
hits.append(self.process(lookup, compound, ind)) |
55
|
|
|
return hits |
56
|
|
|
|
57
|
|
|
def process( |
|
|
|
|
58
|
|
|
self, lookup: str, compound: ChemblCompound, indication: NestedDotDict |
|
|
|
|
59
|
|
|
) -> IndicationHit: |
60
|
|
|
phase = indication.req_as("max_phase_for_ind", int) |
61
|
|
|
return IndicationHit( |
62
|
|
|
record_id=indication.req_as("drugind_id", str), |
63
|
|
|
origin_inchikey=lookup, |
64
|
|
|
matched_inchikey=compound.inchikey, |
65
|
|
|
compound_id=compound.chid, |
66
|
|
|
compound_name=compound.name, |
67
|
|
|
predicate=f"used in a phase {phase} trial for", |
68
|
|
|
object_id=indication.req_as("mesh_id", str), |
69
|
|
|
object_name=indication.req_as("mesh_heading", str).strip("\n"), |
70
|
|
|
search_key=self.key, |
71
|
|
|
search_class=self.search_class, |
72
|
|
|
data_source=self.data_source + " :: " + indication.req_as("references", str), |
73
|
|
|
max_phase=phase, |
74
|
|
|
first_approval_year=indication.req_as("first_approval", str), |
75
|
|
|
) |
76
|
|
|
|
77
|
|
|
|
78
|
|
|
__all__ = ["IndicationHit", "IndicationSearch"] |
79
|
|
|
|