Swallowing the Bitter Pill: Simplified Scalable Conformer Generation
Yuyang Wang Ahmed A. Elhag Navdeep Jaitly Joshua M. Susskind Miguel Ángel Bautista
Abstract
We present a novel way to predict molecular conformers through a simple formulation that sidesteps many of the heuristics of prior works and achieves state of the art results by using the advantages of scale. By training a diffusion generative model directly on 3D atomic positions without making assumptions about the explicit structure of molecules (e.g. modeling torsional angles) we are able to radically simplify structure learning, and make it trivial to scale up the model sizes. This model, called Molecular Conformer Fields (MCF), works by parameterizing conformer structures as functions that map elements from a molecular graph directly to their 3D location in space. This formulation allows us to boil down the essence of structure prediction to learning a distribution over functions. Experimental results show that scaling up the model capacity leads to large gains in generalization performance without enforcing inductive biases like rotational equivariance. MCF represents an advance in extending diffusion models to handle complex scientific problems in a conceptually simple, scalable and effective manner.
中文速览
分子三维构象预测（conformer generation）是药物发现的核心难题，传统方法要么依赖大量人工规则（如扭转角模板），要么用旋转等变网络等专用归纳偏置，导致模型复杂、难以扩展且在复杂分子上容易出错。研究团队提出了"分子构象场"（Molecular Conformer Fields, MCF），将每个分子的构象表示成一个函数——直接把分子图中的原子节点映射到三维空间坐标，然后用扩散生成模型学习这类函数的分布，整个流程不需要预测扭转角、不强制旋转等变性，架构上采用可高效扩展的 PerceiverIO Transformer。在标准构象生成基准上，MCF 超越了此前所有强基线，且随着模型参数量增加，性能提升幅度持续扩大，呈现出明显的"规模红利"。这项工作表明，抛弃领域专用的归纳偏置、转而依靠更大模型从数据中自行学习分子规律，是解决复杂科学问题的一条更简洁、更有潜力的路径。
原文 arXiv:2311.17932;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2311.17932v3