prediction.py

import streamlit as st
import torch
import torch.nn as nn
import numpy as np
import joblib
from transformers import AutoTokenizer, AutoModel
from rdkit import Chem
from rdkit.Chem import Descriptors, AllChem
from datetime import datetime
from db import get_database
import random

# Set seeds
random.seed(42)
np.random.seed(42)
torch.manual_seed(42)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Load ChemBERTa
@st.cache_resource
def load_chemberta():
    tokenizer = AutoTokenizer.from_pretrained("seyonec/ChemBERTa-zinc-base-v1")
    model = AutoModel.from_pretrained("seyonec/ChemBERTa-zinc-base-v1").to(device).eval()
    return tokenizer, model

# Load scalers
scalers = {
    "Tensile Strength(Mpa)": joblib.load("scaler_Tensile_strength_Mpa_.joblib"),
    "Ionization Energy(eV)": joblib.load("scaler_Ionization_Energy_eV_.joblib"),
    "Electron Affinity(eV)": joblib.load("scaler_Electron_Affinity_eV_.joblib"),
    "logP": joblib.load("scaler_LogP.joblib"),
    "Refractive Index": joblib.load("scaler_Refractive_Index.joblib"),
    "Molecular Weight(g/mol)": joblib.load("scaler_Molecular_Weight_g_mol_.joblib")
}

# Transformer model
class TransformerRegressor(nn.Module):
    def __init__(self, feat_dim=2058, embedding_dim=768, ff_dim=1024, num_layers=2, output_dim=6):
        super().__init__()
        self.feat_proj = nn.Linear(feat_dim, embedding_dim)
        encoder_layer = nn.TransformerEncoderLayer(
            d_model=embedding_dim, nhead=8, dim_feedforward=ff_dim,
            dropout=0.1, batch_first=True
        )
        self.transformer_encoder = nn.TransformerEncoder(encoder_layer, num_layers=num_layers)
        self.regression_head = nn.Sequential(
            nn.Linear(embedding_dim, 256),
            nn.ReLU(),
            nn.Linear(256, 128),
            nn.ReLU(),
            nn.Linear(128, output_dim)
        )

    def forward(self, x,feat):
        feat_emb=self.feat_proj(feat)
        stacked=torch.stack([x,feat_emb],dim=1)
        encoded=self.transformer_encoder(stacked)
        aggregated=encoded.mean(dim=1)
        return self.regression_head(aggregated)

# Load model
@st.cache_resource
def load_model():
    model = TransformerRegressor()
    try:
        state_dict = torch.load("transformer_model.bin", map_location=device)
        model.load_state_dict(state_dict)
        model.eval().to(device)
    except Exception as e:
        raise ValueError(f"Failed to load model: {e}")
    return model

# RDKit descriptors
def compute_descriptors(smiles: str):
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError("Invalid SMILES string.")
    desc = [
        Descriptors.MolWt(mol),
        Descriptors.MolLogP(mol),
        Descriptors.TPSA(mol),
        Descriptors.NumRotatableBonds(mol),
        Descriptors.NumHDonors(mol),
        Descriptors.NumHAcceptors(mol),
        Descriptors.FractionCSP3(mol),
        Descriptors.HeavyAtomCount(mol),
        Descriptors.RingCount(mol),
        Descriptors.MolMR(mol)
    ]
    return np.array(desc, dtype=np.float32)

# Morgan fingerprint
def get_morgan_fingerprint(smiles, radius=2, n_bits=2048):
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError("Invalid SMILES string.")
    fp = AllChem.GetMorganFingerprintAsBitVect(mol, radius, nBits=n_bits)
    return np.array(fp, dtype=np.float32).reshape(1,-1)

# ChemBERTa embedding
def get_chemberta_embedding(smiles: str, tokenizer, chemberta):
    inputs = tokenizer(smiles, return_tensors="pt", padding=True, truncation=True)
    with torch.no_grad():
        outputs = chemberta(**inputs)
    return outputs.last_hidden_state.mean(dim=1).to(device)

# Save to DB
def save_to_db(smiles, predictions):
    predictions_clean = {k: float(v) for k, v in predictions.items()}
    doc = {
        "smiles": smiles,
        "predictions": predictions_clean,
        "timestamp": datetime.now()
    }
    db = get_database()
    db["polymer_predictions"].insert_one(doc)

# Streamlit UI
def show():
    st.markdown("<h1 style='text-align: center; color: #4CAF50;'>🔬 Polymer Property Prediction</h1>", unsafe_allow_html=True)
    st.markdown("<hr style='border: 1px solid #ccc;'>", unsafe_allow_html=True)

    smiles_input = st.text_input("Enter SMILES Representation of Polymer")

    if st.button("Predict"):
        try:
            model = load_model()
            tokenizer, chemberta = load_chemberta()

            mol = Chem.MolFromSmiles(smiles_input)
            if mol is None:
                st.error("Invalid SMILES string.")
                return

            descriptors = compute_descriptors(smiles_input)
            descriptors_tensor=torch.tensor(descriptors,dtype=torch.float32).unsqueeze(0)
            fingerprint = get_morgan_fingerprint(smiles_input)
            fingerprint_tensor=torch.tensor(fingerprint,dtype=torch.float32)
            features=torch.cat([descriptors_tensor,fingerprint_tensor],dim=1).to(device)
            embedding = get_chemberta_embedding(smiles_input, tokenizer, chemberta)


            with torch.no_grad():
                preds = model(embedding,features)

            preds_np=preds.cpu().numpy()
            keys = list(scalers.keys())
            preds_rescaled = np.concatenate([
                scalers[keys[i]].inverse_transform(preds_np[:, [i]])
                for i in range(6)
            ], axis=1)

            results = {key: round(val, 4) for key, val in zip(keys, preds_rescaled.flatten())}

            st.success("Predicted Properties:")
            for key, val in results.items():
                st.markdown(f"**{key}**: {val}")

            save_to_db(smiles_input, results)

        except Exception as e:
            st.error(f"Prediction failed: {e}")