Initial commit

data/pattern_penalties.tsv

@@ -0,0 +1,20 @@
+name,penalty
+3_valent_halogen,100
+3_valent_iodine,100
+4_valent_sulphur_2_connections,100
+6_membered_aromatic_sulfur,50
+8_aminoquinoline,100
+8_hydroxyquinoline,100
+9_aminoacridine,100
+acetal_1_in_ring,30
+acetal_acyclic,100
+acetal_both_in_ring,30
+acetate_ester,100
+acetylene,50
+acetylene_heteroatom,100
+acid_halide,100
+activated_ester,100
+activated_phthalimide,100
+acyl_amide_acyclic,40
+acyl_amide_cyclic,60
+acyl_aromatic_nn,100

data/queries/3_valent_halogen.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[#9,#17,#35,#53;X3]"

data/queries/3_valent_iodine.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[#53;X3]"

data/queries/4_valent_sulphur_2_connections.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[SD2v4]"

data/queries/6_membered_aromatic_sulfur.txt

@@ -0,0 +1,2 @@
+SMARTS
+"s1aaaaa1"

data/queries/8_aminoquinoline.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[nD2H0]1c2c(-[ND2H])cccc2ccc1"

data/queries/8_hydroxyquinoline.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[OD1]c1cccc2cccnc12"

data/queries/9_aminoacridine

@@ -0,0 +1,2 @@
+SMARTS
+"N-c1c2c([nH0]c3c1cccc3)cccc2"

data/queries/acetal_1_in_ring.txt

@@ -0,0 +1,2 @@
+SMARTS
+"C[O,S;R1]C[O,S;R0]C"

data/queries/acetal_acyclic.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[#7,#8,#16;X2;!R]-[C;!R;X4]-[#7,#8,#16;X2;!R]"

data/queries/acetal_both_in_ring.txt

@@ -0,0 +1,2 @@
+SMARTS
+C[O,S;R1]C[O,S;R1]C

data/queries/acetate_ester.txt

@@ -0,0 +1,2 @@
+SMARTS	min_hits_needed
+"[#6]-[#8]-[#6;X3](-[#6;H3])=[#8]"	3

data/queries/acetylene.txt

@@ -0,0 +1,2 @@
+SMARTS
+"C#C"

data/queries/acetylene_heteroatom.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[!#6]-C#C"

data/queries/acid_halide.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[#9,#17,#35,#53]-[C;X3]=O"

data/queries/activated_ester.txt

@@ -0,0 +1,2 @@
+SMARTS
+"O=[#6,#15,#16;!R;X3,X4]-O-[#6,#15,#16]"

data/queries/activated_phthalimide.txt

@@ -0,0 +1,3 @@
+SMARTS
+"O=C2c1ccc([F,Cl,Br,I,$(N(~[O;X1])-[O;X1]),$(C=O)])cc1C(=O)N2"
+"O=C2c1cccc([F,Cl,Br,I,$(N(~[O;X1])-[O;X1]),$(C=O)])c1C(=O)N2"

data/queries/acyl_amide_acyclic.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[#8,#16;X1]=[C;!R]-[#7]-[C;!R]=[#8,#16;X1]"

data/queries/acyl_amide_cyclic.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[#8,#16;X1]=[C;R]-[#7]-[C;!R]=[#8,#16;X1]"

data/queries/acyl_aromatic_nn.txt

@@ -0,0 +1,2 @@
+SMARTS
+"[O,S;X1]=[C;!R]-[n;r5;R1;X3]~[n;r5;R1;X2]"

tests/test_smarts.py

@@ -0,0 +1,29 @@
+import os
+from pathlib import Path
+import unittest
+
+import pandas as pd
+from rdkit import Chem
+
+
+class CheckedSmarts(unittest.TestCase):
+    def test_validity(self):
+        data_path = Path(__file__).parents[1] / "data"
+        query_path = data_path / "queries"
+        checked_smarts = 0
+        for file in os.listdir(query_path):
+            smarts_df = pd.read_csv(query_path / file, sep="\t")
+            try:
+                smarts_list = smarts_df["SMARTS"].tolist()
+            except Exception as e:
+                print(file)
+                raise e
+            for smarts_str in smarts_list:
+                smarts_obj = Chem.MolFromSmarts(smarts_str)
+                self.assertIsNotNone(smarts_obj, file)
+            checked_smarts += 1
+        print(checked_smarts)
+
+
+if __name__ == '__main__':
+    unittest.main()