相关实验视频

Updated: Jul 11, 2025

A Facile Protocol to Generate Site-Specifically Acetylated Proteins in Escherichia Coli

A Facile Protocol to Generate Site-Specifically Acetylated Proteins in Escherichia Coli

Published on: December 9, 2017

MSTL-Kace:基于多阶段转移学习策略预测 Prokaryotic Lysine 乙化位

Gang-Ao Wang¹, Xiaodi Yan¹, Xiang Li¹

¹School of Sciences, Anhui Agricultural University, Hefei 230036, Anhui, China.

|November 16, 2023

概括

我们开发了MSTL-Kace,这是一个新的机器学习模型,用于预测 lysine 乙化 (Kace) 位点. 该模型使用转移学习和BERT嵌入来提高准确性,特别是对于数据有限的物种.

科学领域:

生物化学生物化学
生物信息学是一种生物信息学.
计算生物学计算生物学

背景情况:

素乙化 (Kace) 是一种关键的翻译后修饰 (PTM),涉及到各种生物过程.
传统的实验Kace站点识别是艰苦而昂贵的.
现有的机器学习方法经常在特征表示和小样本大小方面扎.

研究的目的:

开发一个准确和高效的计算模型来预测氨酸乙化位点.
为了解决凯斯预测中手工制作的特征和小样本问题的局限性.
为了利用转移学习和深度学习来提高蛋白质修饰部位的预测.

主要方法:

提出了MSTL-Kace,这是一种采用转移学习与预训练的变压器双向编码器表示 (BERT) 的新型模型.
使用特定物种的BERT模型提取的高水平残留物嵌入物.
实施了两阶段的微调策略,以管理小样本数据挑战.
利用随机森林作为下游学习算法,以在六种 prokaryotic 物种上获得最佳性能.

主要成果:

与最先进的方法相比,MSTL-Kace在所有六种 prokaryotic 种类的独立测试集上表现出优越的性能.
该模型通过深度学习嵌入有效地捕获复杂的生物信息.
两阶段微调方法成功缓解了与有限的数据可用性相关的问题.

结论:

MSTL-Kace在计算Kace站点预测方面取得了重大进展.
转移学习策略有效地提高了预测准确性,特别是在代表性不足的物种.
开发的模型为生物研究提供了有价值的工具,需要有效地识别凯斯遗址.

更多相关视频

Quantification of Site-specific Protein Lysine Acetylation and Succinylation Stoichiometry Using Data-independent Acquisition Mass Spectrometry

Quantification of Site-specific Protein Lysine Acetylation and Succinylation Stoichiometry Using Data-independent Acquisition Mass Spectrometry

Published on: April 4, 2018

Site Specific Lysine Acetylation of Histones for Nucleosome Reconstitution using Genetic Code Expansion in Escherichia coli

Site Specific Lysine Acetylation of Histones for Nucleosome Reconstitution using Genetic Code Expansion in Escherichia coli

Published on: December 26, 2020

相关概念视频

Conserved Binding Sites

Conserved Binding Sites

Many proteins’ biological role depends on their interactions with their ligands, small molecules that bind to specific locations on the protein known as ligand-binding sites. Ligand-binding sites are often conserved among homologous proteins as these sites are critical for protein function.
Binding sites are often located in large pockets, and if their location on a protein’s surface is unknown, it can be predicted using various approaches. The energetic method computationally...