From In Silico to In Vitro: Evaluating Molecule Generative Models for Hit Generation

2025-12-26

AI Summary

High-throughput screening (HTS) remains costly and time-consuming in early drug discovery; this work investigates whether generative AI can directly generate hit-like molecules—bypassing HTS entirely. Method: We formally define hit-like molecule generation as a distinct task and introduce a multi-stage evaluation framework integrating rational property optimization, structural diversity, and target-specific bioactivity (e.g., experimental validation against GSK-3β *in vitro*). We systematically benchmark two autoregressive models and one diffusion model across diverse datasets and training regimes, employing standard generative metrics alongside molecular docking scores. Contribution/Results: Generated molecules exhibit high validity, structural diversity, and biological relevance. Multiple *de novo* predicted hits were synthesized and experimentally confirmed to possess significant inhibitory activity. The study further identifies critical limitations of current evaluation metrics and training data biases, establishing a rigorously validated, AI-driven paradigm for hit generation.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Hit identification is a critical yet resource-intensive step in the drug discovery pipeline, traditionally relying on high-throughput screening of large compound libraries. Despite advancements in virtual screening, these methods remain time-consuming and costly. Recent progress in deep learning has enabled the development of generative models capable of learning complex molecular representations and generating novel compounds de novo. However, using ML to replace the entire drug-discovery pipeline is highly challenging. In this work, we rather investigate whether generative models can replace one step of the pipeline: hit-like molecule generation. To the best of our knowledge, this is the first study to explicitly frame hit-like molecule generation as a standalone task and empirically test whether generative models can directly support this stage of the drug discovery pipeline. Specifically, we investigate if such models can be trained to generate hit-like molecules, enabling direct incorporation into, or even substitution of, traditional hit identification workflows. We propose an evaluation framework tailored to this task, integrating physicochemical, structural, and bioactivity-related criteria within a multi-stage filtering pipeline that defines the hit-like chemical space. Two autoregressive and one diffusion-based generative models were benchmarked across various datasets and training settings, with outputs assessed using standard metrics and target-specific docking scores. Our results show that these models can generate valid, diverse, and biologically relevant compounds across multiple targets, with a few selected GSK-3$β$ hits synthesized and confirmed active in vitro. We also identify key limitations in current evaluation metrics and available training data.

Problem

• Evaluating generative models for hit-like molecule generation
• Proposing a tailored evaluation framework for hit identification
• Benchmarking models across datasets with synthesis validation

Innovation

• Generative models replace hit-like molecule generation step
• Multi-stage filtering pipeline defines hit-like chemical space
• Benchmark autoregressive and diffusion models with docking scores