From 7b836ed4d5bedbc478b6fd91f880791ff488bfc0 Mon Sep 17 00:00:00 2001
From: Kohulan <kohulan.rajan@uni-jena.de>
Date: Tue, 7 Mar 2023 19:17:28 +0100
Subject: [PATCH] feat: descriptor calculations using RDKit

---
 app/modules/descriptor_calculator.py | 64 ++++++++++++++++++++++++++++
 app/routers/chem.py                  |  4 +-
 2 files changed, 66 insertions(+), 2 deletions(-)
 create mode 100644 app/modules/descriptor_calculator.py

diff --git a/app/modules/descriptor_calculator.py b/app/modules/descriptor_calculator.py
new file mode 100644
index 0000000..e5cda3c
--- /dev/null
+++ b/app/modules/descriptor_calculator.py
@@ -0,0 +1,64 @@
+from rdkit import Chem
+from rdkit.Chem import Descriptors, QED, Lipinski, rdMolDescriptors, rdmolops
+
+
+def checkRo5Violations(mol):
+    """Takes a molecules and checks whether the molecule violates
+    the Lipinski's Rule of Five.
+    Args : molecules rdkit.Chem.rdmol.Mol: rdkit_mol Objects
+    Returns (int): A number of violations on Lipinski Rules.
+    """
+    num_of_violations = 0
+    if Descriptors.MolLogP(mol) > 5:
+        rule_break += 1
+    if Descriptors.MolWt(mol) > 500:
+        rule_break += 1
+    if Lipinski.NumHAcceptors(mol) > 10:
+        rule_break += 1
+    if Lipinski.NumHDonors(mol) > 5:
+        rule_break += 1
+    return num_of_violations
+
+
+def GetBasicDescriptors(smiles):
+    """Take an input SMILES and generates a selected set of molecular
+    descriptors as a dictionary
+    Args (str): SMILES string
+    Returns (dict): a dictionary of calculated descriptors
+    """
+    mol = Chem.MolFromSmiles(smiles)
+    AtomC = rdMolDescriptors.CalcNumAtoms(mol)
+    HeavyAtomsC = rdMolDescriptors.CalcNumHeavyAtoms(mol)
+    MolWt = "%.2f" % Descriptors.MolWt(mol)
+    ExactMolWt = "%.2f" % Descriptors.ExactMolWt(mol)
+    ALogP = "%.2f" % QED.properties(mol).ALOGP
+    NumRotatableBonds = rdMolDescriptors.CalcNumRotatableBonds(mol)
+    PSA = "%.2f" % rdMolDescriptors.CalcTPSA(mol)
+    HBA = Descriptors.NumHAcceptors(mol)
+    HBD = Descriptors.NumHDonors(mol)
+    Lipinski_HBA = Lipinski.NumHAcceptors(mol)
+    Lipinski_HBD = Lipinski.NumHDonors(mol)
+    Ro5Violations = checkRo5Violations(mol)
+    AromaticRings = rdMolDescriptors.CalcNumAromaticRings(mol)
+    QEDWeighted = "%.2f" % QED.qed(mol)
+    FormalCharge = "%.2f" % rdmolops.GetFormalCharge(mol)
+
+    AllDescriptors = {
+        "Atom count": AtomC,
+        "Heavy atom count": HeavyAtomsC,
+        "Molecular weight": MolWt,
+        "Exact molecular weight": ExactMolWt,
+        "ALogP": ALogP,
+        "Rotatable bond count": NumRotatableBonds,
+        "Topological polar surface area": PSA,
+        "Hydrogen bond acceptors": HBA,
+        "Hydrogen bond donors": HBD,
+        "Hydrogen bond acceptors(Lipinski)": Lipinski_HBA,
+        "Hydrogen bond donors(Lipinski)": Lipinski_HBD,
+        "Lipinski's rule of five violations": Ro5Violations,
+        "Aromatic rings count": AromaticRings,
+        "QED drug likeliness": QEDWeighted,
+        "Formal Charge": FormalCharge,
+    }
+
+    return AllDescriptors
diff --git a/app/routers/chem.py b/app/routers/chem.py
index 22c005d..19c350b 100644
--- a/app/routers/chem.py
+++ b/app/routers/chem.py
@@ -15,6 +15,7 @@
 from rdkit.Chem.rdMolDescriptors import Properties
 from STOUT import translate_forward, translate_reverse
 from app.modules.npscorer import getnp_score
+from app.modules.descriptor_calculator import GetBasicDescriptors
 
 router = APIRouter(
     prefix="/chem",
@@ -96,8 +97,7 @@ async def standardise_mol(request: Request):
 @router.get("/{smiles}/descriptors")
 async def smiles_descriptors(smiles: Optional[str]):
     if smiles:
-        m = Chem.MolFromSmiles(smiles)
-        return properties(m)
+        return GetBasicDescriptors(smiles)
 
 
 @router.get("/{smiles}/iupac")