聚合残留水平蛋白质语言模型嵌入与最佳的运输
Navid NaderiAlizadeh1, Rohit Singh1,2
1Department of Biostatistics and Bioinformatics, Duke University, Durham, NC 27705, United States.
Bioinformatics advances
|April 2, 2025
概括
蛋白质语言模型 (PLM) 生成可变长度嵌入,对下游任务构成挑战. 我们的新型最佳运输方法创建了固定长度的蛋白质表示,优于传统的聚合方法.
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 机器学习
背景情况:
- 蛋白质语言模型 (PLM) 从蛋白质序列中生成每残留嵌入.
- 来自PLM的可变长度输出阻碍了需要统一输入的蛋白质水平预测任务.
- 平均聚合是一种常见的,但潜在的低于最佳的方法来总结PLM输出.
研究的目的:
- 开发一种用于将可变长度的PLM嵌入式转换为固定长度表示的新方法.
- 为了应对蛋白质表示学习中可变序列长度的挑战.
- 通过使用PLM来提高蛋白质水平预测任务的性能.
主要方法:
- 利用最佳运输理论来汇总每个token的PLM输出.
- 概念化代币嵌入作为概率分布的样本.
- 采用切片的瓦瑟斯坦距离来创建固定长度的蛋白质水平嵌入.
主要成果:
- 建议的最佳运输方法产生了独立于蛋白质长度的固定长度嵌入.
- 这种方法在各种下游任务中比平均聚合显示出更高的性能.
- 使用这种方法的较小的PLM实现了与使用平均聚合的较大PLM可比的性能,特别是在更长的序列中.
结论:
- 最佳运输提供了一个有效的策略,用于从PLM中生成固定长度的蛋白质嵌入.
- 该方法增强了PLM用于蛋白质水平预测的实用性,特别是对于长序列.
- 这种方法可以更有效地使用较小的PLM,同时保持高性能.
相关概念视频
Cotranslational Protein Translocation
7.0K
Translocation of proteins across membranes is an ancient process that occurs even in bacteria and archaebacteria. In fact, the components of the translocation machinery are still conserved between prokaryotes and eukaryotes.
Sec61 channel partners for cotranslational translocation
During cotranslational translocation, the Sec61 channel partners with the signal recognition particle (SRP), the signal recognition particle receptor (SR), and the ribosomes to transport the nascent polypeptide chain...
Sec61 channel partners for cotranslational translocation
During cotranslational translocation, the Sec61 channel partners with the signal recognition particle (SRP), the signal recognition particle receptor (SR), and the ribosomes to transport the nascent polypeptide chain...
7.0K
Translocation of Proteins into the Mitochondria
3.0K
Mitochondrial precursors are translocated to the internal subcompartments via independent mechanisms involving distinct protein machineries called translocases.
Sorting of outer membrane proteins:
Mitochondrial outer membrane proteins are of two types: the transmembrane, beta-barrel porins, and the membrane-anchored, alpha-helical proteins. Beta-barrel porin precursors are translocated by the TOM complex and inserted into the outer mitochondrial membrane by the SAM complex. In contrast,...
Sorting of outer membrane proteins:
Mitochondrial outer membrane proteins are of two types: the transmembrane, beta-barrel porins, and the membrane-anchored, alpha-helical proteins. Beta-barrel porin precursors are translocated by the TOM complex and inserted into the outer mitochondrial membrane by the SAM complex. In contrast,...
3.0K
Transport Across the Golgi
4.0K
While it is unclear how molecules move between adjacent Golgi cisternae, it is apparent that the molecules move from cis- cisterna, the entry face, to the trans- cisterna, the exit face. Experiments initially suggested vesicles that bud from one cisterna and fuse with the next cisterna to transport proteins between the cisternae. This vesicular transport model describes the Golgi apparatus as a relatively static structure with a unique enzyme composition in each cisterna. Molecules are...
4.0K
Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving
34
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...
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...
34


