相关实验视频

Updated: May 27, 2025

Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches

Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches

Published on: December 15, 2023

从机械解释性到机械生物学:在蛋白质语言模型上训练,评估和解释稀缺的自编码器

Etowah Adams¹, Liam Bai², Minji Lee¹

¹Department of Systems Biology, Columbia University.

bioRxiv : the preprint server for biology

|February 20, 2025

概括

蛋白质语言模型 (pLMs) 使用通用和家族特定特征表示蛋白质. 使用稀疏自编码器 (SAEs) 分析这些特征可以揭示新的生物机制,并改善pLM的理解.

科学领域:

计算生物学计算生物学
生物信息学是一种生物信息学.
机器学习在生物学中的应用

背景情况:

蛋白质语言模型 (pLMs) 是用于预测蛋白质结构和功能的先进工具,在庞大的序列数据集上进行训练.
由pLMs学习的内部特征或表示并不完全理解,限制了对其预测能力和生物发现潜力的洞察力.
了解pLM特征可以揭示这些模型如何捕获生物信息,并可能揭示新的蛋白质生物学.

研究的目的:

为了研究蛋白质语言模型 (pLMs) 学习的特定特征.
描述这些特征的性质 (通用与特定) 以及它们与蛋白质性质的关系.
探索这些特征对于产生关于未知的生物机制的假设的实用性.

主要方法:

在ESM-2 pLM的剩余流输出上培训稀疏自动编码器 (SAE).
描述已学习的SAE特征,以了解它们对蛋白质序列信息的表示.
使用SAE特征的线性探测来识别蛋白质性质的序列决定因素,如热稳定性和局部化.
开发可视化工具来解释预测SAE特征.

主要成果:

证明了pLMs利用一系列的通用和家族特征来代表蛋白质.
通过SAE特征的线性探测成功确定了蛋白质热稳定性和亚细胞局部化的已知序列决定因素.
突出预测特征缺乏明确的功能关联,表明在未被发现的生物机制中的潜在作用.

结论:

pLM 特性提供了一般和特定生物信息的混合.
SAE分析提供了一种方法来解释pLM表示,并将它们与蛋白质功能联系起来.
这种方法提高了对pLM局限性的理解,并促进了新的生物假设的产生.

更多相关视频

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

A Protocol for Computer-Based Protein Structure and Function Prediction

A Protocol for Computer-Based Protein Structure and Function Prediction

Published on: November 3, 2011

相关概念视频

Conservation of Protein Domains Over Different Proteins

Conservation of Protein Domains Over Different Proteins

Protein domains are small structurally independent units that are part of a single amino acid chain. Although these domains are often structurally independent, they may rely on synergistic effects to perform their functions as part of a larger protein. Protein domains may be conserved within the same organism, as well as across different organisms.
A limited set of protein domains often duplicate and recombine during evolution. These domains can be organized in different combinations to...

Protein Organization

Protein Organization

Proteins are polymers of amino acid residues. They are versatile and responsible for different cellular functions, including DNA replication, molecular transport, catalysis, and structural support. Proteins have a hierarchical structure comprising at least three levels of organization: primary, secondary, and tertiary structure. Some large proteins have a quaternary structure where individual protein subunits are linked together.
The primary structure of a protein is its amino acid sequence....

Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving

Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving

Mechanistic models play a crucial role in algorithms for numerical problem-solving, particularly in nonlinear mixed effects modeling (NMEM). These models aim to minimize specific objective functions by evaluating various parameter estimates, leading to the development of systematic algorithms. In some cases, linearization techniques approximate the model using linear equations.
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...

Leaky Scanning

Leaky Scanning

During most eukaryotic translation processes, the small 40S ribosome subunit scans an mRNA from its 5' end until it encounters the first start AUG codon. The large 60S ribosomal subunit then joins the smaller one to initiate protein synthesis. The location of the translation initiation is largely determined by the nucleotides near the start codon as there may be multiple translation initiation sites present on the mRNA. Marilyn Kozak discovered that the sequence RCCAUGG (where R...