there and back again

README.md

@@ -26,7 +26,7 @@ You can try ugropy from its
 [Binder](https://mybinder.org/v2/gh/ipqa-research/ugropy/main). Open the
 binder.ipynb file to explore the basic features.
 
-# Models supported v2.0.7
+# Models implemented
 - Classic liquid-vapor UNIFAC
 - Predictive Soave-Redlich-Kwong (PSRK)
 - Joback

pyproject.toml

@@ -35,7 +35,7 @@ classifiers = [
 urls = {Homepage = "https://github.com/ipqa-research/ugropy"}
 
 dependencies = [
-    'numpy >= 1.25.1, <2.0.0',
+    'numpy >= 1.25.1',
     'pandas >= 2.0.3',
     'pubchempy >= 1.0.4',
     'rdkit >= 2023.9.5'

requirements-dev.txt

@@ -26,9 +26,3 @@ ipdb
 ipython
 jupyter
 matplotlib
-
-# ugropy dependecies
-numpy >= 1.25.1, <2.0.0
-pandas >= 2.0.3
-pubchempy >= 1.0.4
-rdkit >= 2023.9.5

ugropy/refactor/__init__.py

@@ -0,0 +1,6 @@
+from .fragment import Fragment
+from .fragmentation_model import FragmentationModel
+from .fragmentation_unifac import unifac
+
+
+__all__ = ["Fragment", "FragmentationModel", "unifac"]

ugropy/refactor/fragment.py

@@ -0,0 +1,8 @@
+from rdkit import Chem
+
+
+class Fragment:
+    def __init__(self, name: str, smarts: str):
+        self.name = name
+        self.smarts = smarts
+        self.mol_object = Chem.MolFromSmarts(smarts)

ugropy/refactor/fragmentation_model.py

@@ -0,0 +1,21 @@
+from typing import List
+
+from ugropy.refactor.fragment import Fragment
+
+from rdkit import Chem
+
+
+class FragmentationModel:
+    def __init__(self, fragments: List[Fragment]):
+        self.fragments = fragments
+
+    def detect_fragments(self, molecule: Chem.rdchem.Mol):
+        detected = {}
+
+        for fragment in self.fragments:
+            match = molecule.GetSubstructMatches(fragment.mol_object)
+
+            if match:
+                detected[fragment.name] = match
+
+        return detected

ugropy/refactor/fragmentation_unifac.py

@@ -0,0 +1,22 @@
+from ugropy.refactor.fragment import Fragment
+from ugropy.refactor.fragmentation_model import FragmentationModel
+
+
+_unifac_fragments = [
+    Fragment("CH3", "[CH3]"),
+    Fragment("CH2", "[CH2]"),
+    Fragment("CH", "[CH]"),
+    Fragment("C", "[CH0]"),
+    Fragment("OH", "[OH]"),
+    Fragment("H2O", "O"),
+    Fragment("ACH", "[cH]"),
+    Fragment("AC", "[cH0]"),
+    Fragment("ACCH3", "[cH0][CH3]"),
+    Fragment("ACCH2", "[cH0][CH2]"),
+    Fragment("ACCH", "[cH0][CH]"),
+]
+
+
+unifac = FragmentationModel(_unifac_fragments)
+
+# mol = Chem.MolFromSmiles("C(C1=CC=CC=C1)C1=CC=CC=C1")