Generative Enriched Sequential Learning (ESL) Approach for Molecular Design via Augmented Domain Knowledge

Abstract

Deploying generative machine learning techniques to generate novel chemical structures based on molecular fingerprint representation has been well established in molecular design. Typically, sequential learning (SL) schemes such as hidden Markov models (HMM) and, more recently, in the sequential deep learning context, recurrent neural network (RNN) and long short-term memory (LSTM) were used extensively as generative models to discover unprecedented molecules. To this end, emission probability between two states of atoms plays a central role without considering specific chemical or physical properties. Lack of supervised domain knowledge can mislead the learning procedure to be relatively biased to the prevalent molecules observed in the training data that are not necessarily of interest. We alleviated this drawback by augmenting the training data with domain knowledge, e.g. quantitative estimates of the drug-likeness score (QEDs). As such, our experiments demonstrated that with this subtle trick called enriched sequential learning (ESL), specific patterns of particular interest can be learnt better, which led to generating de novo molecules with ameliorated QEDs.

Machine learning for the prediction of safe and biologically active organophosphorus molecules

“…Over the past decade, machine learning-based de novo molecule designs have gained increased popularity and have been applied in various systems [14,15]. One common approach is to adopt natural language processing methods, such as recurrent neural networks (RNN) with self-attention have shown the ability to overcome typical energy space barriers and predict novel molecules with good chemical sense.…”

“…One common approach is to adopt natural language processing methods, such as recurrent neural networks (RNN) with self-attention have shown the ability to overcome typical energy space barriers and predict novel molecules with good chemical sense. Part of RNN's popularity can be attributed to the LSTM (long short-term memory) developed in 1997 [15,16]. LSTM cell helps RNN consider the input sequence's long-and short-term dependency.…”

ChemSpaceAL: An Efficient Active Learning Methodology Applied to Protein-Specific Molecular Generation

“…Virtual screening and de novo drug design are popular areas of research aimed at developing effective protein-specific drugs . Molecular generation methods powered by generative artificial intelligence (AI) can advance both of these areas, and there have already been numerous reports of recurrent neural networks (RNNs), − generative adversarial networks (GANs), − autoencoders, − and transformers − successfully contributing to drug development methods.…”