MolLM:一种统一的语言模型,用于将生物医学文本与2D和3D分子表示集成在一起
Xiangru Tang1, Andrew Tran1, Jeffrey Tan1
1Department of Biomedical Informatics & Data Science, Yale University, New Haven, CT 06520, USA.
Bioinformatics (Oxford, England)
|June 28, 2024
概括
这项研究介绍了MolLM,一种新的语言模型,将2D和3D分子结构与生物医学文本集成在一起. 整合3D信息可以在各种下游任务中增强分子表示和性能.
科学领域:
- 计算化学计算化学
- 生物信息学是一种生物信息学.
- 人工智能的人工智能
背景情况:
- 目前对分子的深度学习模型经常忽略3D结构数据,限制了它们的实用性.
- 现有的以3D为重点的方法可能会忽略关键的生物医学文本信息.
- 需要采用统一的方法来整合各种分子和文本数据.
研究的目的:
- 开发一个统一的预训练语言模型,MolLM,能够与生物医学文本共同表示2D和3D分子结构.
- 为了利用自我监督的学习与强大的分子和文本嵌入的对比目标.
- 评估MolLM在各种下游任务上的表现,包括交叉模式匹配,属性预测和分子编辑.
主要方法:
- 开发了MolLM,该模型为文本和2D/3D分子结构提供了单独的变压器编码器.
- 创建了一个由16万个分子-文本配对组成的数据集,用于自我监督的预训.
- 利用对比式学习来训练模型对分子-文本配对.
- 在四个下游任务中评估了MolLM:分子-文本匹配,属性预测,标题和文本提示的分子编辑.
主要成果:
- MolLM成功地将2D和3D分子信息与生物医学文本集成在一起.
- 该模型在所有四个评估的下游任务上都表现出强的表现.
- 废弃研究证实,包含明确的3D分子表示显著提高了性能.
- 该模型实现了强大的分子表示能力.
结论:
- 统一的MolLM框架有效地捕获分子和文本的多模式信息.
- 明确的3D结构信息对于增强分子表示学习至关重要.
- MolLM为各种生物医学和化学信息应用提供了多功能基础.
相关概念视频
02:00Molecular Models
38.2K
Physical models representing molecular architectures of chemical compounds play essential roles in understanding chemistry. The use of molecular models makes it easier to visualize the structures and shapes of atoms and molecules.
38.2K
01:16Calmodulin-dependent Signaling
5.1K
Calmodulin (CaM) is a calcium-binding protein in eukaryotes that controls various calcium-regulated cellular processes. It has four calcium-binding sites that bind calcium to form the calcium-calmodulin ( Ca2+-CaM) complex. GPCR stimulation increases the calcium levels in the cells that bind to CaM and induces a conformational change.
The Ca2+-CaM complex does not have enzymatic activity by itself. Instead, the complex binds downstream target proteins, including membrane proteins or enzymes,...
The Ca2+-CaM complex does not have enzymatic activity by itself. Instead, the complex binds downstream target proteins, including membrane proteins or enzymes,...
5.1K
02:21Globular and Fibrous Proteins
43.6K
Many proteins can be classified into two distinct subtypes - globular or fibrous. These two types differ in their shapes and solubilities.
Globular proteins are also known as spheroproteins and typically are approximately round in shape. They contain a mix of amino acid types and contain differing sequences in their primary structures. Globular proteins have many different functions, such as enzymes, cellular messengers, and molecular transporters. These roles often require the proteins to be...
Globular proteins are also known as spheroproteins and typically are approximately round in shape. They contain a mix of amino acid types and contain differing sequences in their primary structures. Globular proteins have many different functions, such as enzymes, cellular messengers, and molecular transporters. These roles often require the proteins to be...
43.6K
01:57Gene Families
8.8K
Gene families consist of groups of genes proposed to have originated from a common ancestor. Typically these arise through events in which a gene or genes are mistakenly duplicated during cell division. Unlike their parent genes (which are subject to selection pressure to maintain function), these gene copies do not need to preserve their sequences and may evolve at a relatively faster rate.
Occasionally these regions can be adapted to take on new roles within the organism, becoming novel genes...
Occasionally these regions can be adapted to take on new roles within the organism, becoming novel genes...
8.8K