A combined method based on wavelet transform for signal Peptide Local Features Recognition
会议名称：《第九届全国计算（机）化学学术会议》
会议日期：2007年
学科分类：081704[工学-应用化学] 07[理学] 08[工学] 0817[工学-化学工程与技术] 070303[理学-有机化学] 0703[理学-化学]
关 键 词：Signal Peptide Cleavage Site Wavelet Transform
摘 要：正Signal peptides direct nascent proteins to certain organelles and are cleaved off after the proteins *** the rapid increase of sequence entries,it is highly desirable to develop automated algorithm to identify the signal peptide or predict its cleavage ***,many methods have been *** vast majority of existing predictive methods resorted to a fixed window pattern,*** example,weight-matrix,an early computational method for predicting cleavage sites,employed a so-called summation window of（-13,+2）to predict cleavage sites of both eukaryotic and prokaryotic.A predictor SignalP was developed for both signal peptide and its cleavage site based on neural *** applied different window patterns to five datasets:（-15, 4-4）to human,（-17,+2）to eukaryote,（-11,+2）to Gram-negative bacteria,（-21,+2）to Gram-positive bacteria and（-15,+2）to *** prompts us to propose a model that is universal for all signal peptides. The properties of signal peptides are known to differ between various types of organisms, including the length and *** has been reported that the location of cleavage sites are related to several local featUres in signal *** this context,studies on these local features might be helpful for constructing such a model and,even,for the further researches on signal *** construct such a model,it shall firstly encounter the problem that signal peptides are various in different organism *** a solution to this,the architecture of the model is separated into two *** the first step,h-region is *** this,the last three characteristic amino acids（Here,we call it（-3,-1）segment）of each signal peptide is *** the submodel of detecting h-region,a strategy of two-reconstructed profiles with wavelet transform was *** is well performed and demonstrated not affected by the variety of h-regious. This combined model is convenient to different organism *** predicting h-re