Statistical Mechanical Approach to Protein Folding Problem

我们如何才能从大量寻找能够从所有可能的序列中折叠成独特结构的多肽中解脱出来？使用计算机模型，从随机多肽中进行连续的功能选择，使得多肽的给定活性位点进化为经常采取结合活性所需的构型。在200代的时间里，只有在活性位点的局部构型受到限制的情况下，整体蛋白质样结构才逐渐沿着可折叠性、螺旋性和致密性进化。在15个不同的初始随机序列中，有5个进化出了不同的蛋白质样结构，这表明很大一部分序列可以在很短的时间内具有蛋白质样结构。因此，局部构型上的功能选择使得在序列空间中对组织良好的蛋白质样结构的穷举搜索变得容易。

How can we be relieved from the massive search for polypeptides capable of folding into unique structures from all possible sequences? Using a computer model, the successive functional selections from random polypeptides were carried out such that the given active sites of the polypeptide evolved to frequently take the configuration required for a binding activity. Only with the constraint on the local configuration of the active sites, global protein-like structures gradually evolved along with foldability, helicity and compactness within 200 generations. Out of the 15 different initial random sequences five evolved to give different protein-like structures, indicating that a significant fraction of sequences can have the protein-like structures within a short span. Therefore, functional selection on local configuration alleviates the exhaustive search for well-organized protein-like structures in the sequence space.