Skip to content
The Chat Archive
Human-written demonstration (OpenAssistant volunteers)· OpenAssistant (human-written) OpenAssistant Conversations (oasst2)

Write a python script to load a list of SMILES or SMARTS strings from a csv file. Clean…

In OpenAssistant data, “assistant” turns were written by human volunteers as demonstrations. They are not model outputs.
  1. User

    Write a python script to load a list of SMILES or SMARTS strings from a csv file. Clean the input data and remove any duplicates or tautomers. Make sure all protonation states are appropriate for pH 7. Create a small LSTM to predict the label found in the csv file.

  2. Assistant

    Here is a basic implementation of the script to load the SMILES or SMARTS strings from a csv file and clean the data before feeding it into a small LSTM model for prediction:

    import pandas as pd import numpy as np import rdkit from rdkit import Chem from rdkit.Chem import Draw from rdkit.Chem import Descriptors from rdkit.Chem import rdmolops from rdkit.Chem import AllChem

    Load the data from the csv file

    df = pd.read_csv('smiles_data.csv')

    Remove duplicates and tautomers

    df = df.drop_duplicates() df = df.apply(rdmolops.RemoveHs, axis=1)

    Ensure all protonation states are appropriate for pH 7

    df = df.apply(rdmolops.AssignStereochemistry, axis=1) df = df.apply(rdmolops.AssignAtomChiralTagsFromStructure, axis=1) df = df.apply(rdmolops.AssignBondStereochemistry, axis=1)

    Convert the SMILES or SMARTS strings to RDKit molecules

    df['Molecules'] = df['SMILES or SMARTS'].apply(Chem.MolFromSmiles)

    Split the data into training and testing sets

    train_df = df.sample(frac=0.8, random_state=0) test_df = df.drop(train_df.index)

    Create a small LSTM to predict the label

    import tensorflow as tf from tensorflow import keras

    Preprocessing

    def create_dataset(df): data = df['SMILES or SMARTS'].to_numpy() labels = df['Labels'].to_numpy() return data, labels

    train_data, train_labels = create_dataset(train_df) test_data, test_labels = create_dataset(test_df)

    Encoding

    tokenizer = keras.preprocessing.text.Tokenizer(num_words=20000) tokenizer.fit_on_texts(train_data)

    train_encoded = tokenizer.texts_to_sequences(train_data) test_encoded = tokenizer.texts_to_sequences(test_data)

    Padding

    train_padded = keras.preprocessing.sequence.pad_sequences(train_encoded, maxlen=100) test_padded = keras.preprocessing.sequence.pad_sequences(test_encoded, maxlen=100)

    Building the Model

    model = keras.Sequential([ keras.layers.Embedding(20000, 64, input_length=100), keras.layers.Bidirectional(keras.layers.LSTM(64)), keras.layers.Dense(64, activation='relu'), keras.layers.Dense(1, activation='sigmoid') ])

    model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

    history = model.fit

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.