Update app.py
Browse files
app.py
CHANGED
@@ -1,3 +1,42 @@
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from rdkit import Chem
|
2 |
+
from rdkit.Chem import Draw
|
3 |
+
from transformers import pipeline
|
4 |
import gradio as gr
|
5 |
|
6 |
+
model_checkpoint = "yzimmermann/FART"
|
7 |
+
classifier = pipeline("text-classification", model=model_checkpoint, return_all_scores=True)
|
8 |
+
|
9 |
+
def process_smiles(smiles):
|
10 |
+
# Validate and canonicalize SMILES
|
11 |
+
mol = Chem.MolFromSmiles(smiles)
|
12 |
+
if mol is None:
|
13 |
+
return "Invalid SMILES", None, "Invalid SMILES"
|
14 |
+
canonical_smiles = Chem.MolToSmiles(mol)
|
15 |
+
|
16 |
+
# Predict using the pipeline
|
17 |
+
predictions = classifier(canonical_smiles)
|
18 |
+
|
19 |
+
# Generate molecule image
|
20 |
+
img_path = "molecule.png"
|
21 |
+
img = Draw.MolToImage(mol)
|
22 |
+
img.save(img_path)
|
23 |
+
|
24 |
+
# Convert predictions to a friendly format
|
25 |
+
prediction_dict = {pred["label"]: pred["score"] for pred in predictions[0]}
|
26 |
+
|
27 |
+
return prediction_dict, img_path, canonical_smiles
|
28 |
+
|
29 |
+
# Set up the Gradio interface
|
30 |
+
iface = gr.Interface(
|
31 |
+
fn=process_smiles,
|
32 |
+
inputs=gr.inputs.Textbox(label="Input SMILES"),
|
33 |
+
outputs=[
|
34 |
+
gr.outputs.Label(num_top_classes=3, label="Classification Probabilities"),
|
35 |
+
gr.outputs.Image(type="file", label="Molecule Image"),
|
36 |
+
gr.outputs.Textbox(label="Canonical SMILES")
|
37 |
+
],
|
38 |
+
title="FART",
|
39 |
+
description="Enter a SMILES string to get the taste classification probabilities."
|
40 |
+
)
|
41 |
+
|
42 |
+
iface.launch()
|