Applicability of Computational Language Technologies to Identify Independent Protein Folding Domains in Human Proteins

一些证据表明，自然语言和生物序列之间存在着强烈的相似性，即，在完全测序的生物体基因组编码的蛋白质集体中，似乎存在着生物体特有的单词、短语和段落。本文提出，自然文本中意义的生物学类比是蛋白质序列折叠成其功能性三维折叠的能力。在自然语言中，频繁的词没有什么意义，而罕见的词通常可以识别特定文本的主题。因此，该假说预测，氨基酸的罕见延伸表明折叠引发剂的位置。溶菌酶是蛋白质折叠研究的一种模式蛋白，其整体特性沿序列的分布表明，当绘制氨基酸n-gram的逆频率时，可以识别这些特性的特征，支持与自然语言的类比。在接下来的12个月里，重点将放在研究人类基因组中罕见的n-克的分布上。如果折叠结构域和稀有n-grams分布之间的相关性能够建立起来，这将(i)为这一假设提供强有力的支持，（ii）阐明当今生物学中一个主要的未解决问题，即由一维氨基酸序列形成功能性三维结构的机制。

Several lines of evidence suggest that there exists a strong analogy between natural languages and biological sequences, i.e. there appear to be organism-specific words, phrases and paragraphs in the collective of proteins encoded by the genomes of fully sequenced organisms. It is proposed that the biological analogy of meaning in a natural text is the ability of a protein sequence to fold into its functional three-dimensional fold. In natural languages, frequent words carry little meaning, while rare words often allow identification of the topic of a particular text. The hypothesis predicts, therefore, that rare stretches of amino acids indicate the location of folding initiators. The distribution of global properties along the sequence of lysozyme, a model protein for protein folding studies, indicated that features in these properties can be recognized when inverse frequencies of amino acid n-grams were plotted, supporting the analogy to natural languages. In the next 12 months the focus will be on studying the distribution of rare n-grams in the human genome. If a correlation between folding domains and distribution of rare n-grams can be established, this would (i) provide compelling support for the hypothesis and (ii) shed light on one of the major unsolved questions in biology today, the mechanism by which functional three-dimensional structures are formed from a one-dimensional sequence of amino acids.