iterb-PPse: Identification of transcriptional terminators in bacterial by incorporating nucleotide properties into PseKNC

Abstract

23Terminator is a DNA sequence that give the RNA polymerase the transcriptional 24 termination signal. Identifying terminators correctly can optimize the genome 25 annotation, more importantly, it has considerable application value in disease diagnosis 26 and therapies. However, accurate prediction methods are deficient and in urgent need.27 Therefore, we proposed a prediction method "iterb-PPse" for terminators by 28 incorporating 47 nucleotide properties into PseKNC-Ⅰ and PseKNC-Ⅱ and utilizing 29 Extreme Gradient Boosting to predict terminators based on Escherichia coli and 30 Bacillus subtilis. Combing with the preceding methods, we employed three new feature 31 extraction methods K-pwm, Base-content, Nucleotidepro to formulate raw samples. 32 The two-step method was applied to select features. When identifying terminators 33 based on optimized features, we compared five single models as well as 16 ensemble 34 models. As a result, the accuracy of our method on benchmark dataset achieved 35 99.88%, higher than the existing state-of-the-art predictor iTerm-PseKNC in 100 times 36 five-fold cross-validation test. It's prediction accuracy for two independent datasets 37 reached 94.24% and 99.45% respectively. For the convenience of users, a software was 38 developed with the same name on the basis of "iterb-PPse". The open software and 39 source code of "iterb-PPse" are available at https://github.com/Sarahyouzi/iterb-PPse. 3 40 1 Introduction 41 DNA transcription is an important step in the inheritance of genetic information 42 and terminators control the termination of transcription which exists in sequences that 43 have been transcribed. When transcription, the terminator will give the RNA 44 polymerase the transcriptional termination signal. Identifying terminators accurately 45 can optimize the genome annotation, more importantly, it has great application value 46 in disease diagnosis and therapies, so it is crucial to identify terminators. Whereas, 47 using traditional biological experiments to identify terminators is extremely time 48 consuming and labor intensive. Therefore, a more effective and convenient began to be 49 applied in researches, that is, adopting machine learning to identify gene sequences. 50 Previous research found there are two types of terminators in prokaryotes, namely 51 Rho-dependent and Rho-independent[1], as shown in Fig 1. Although there have been 52 a lot of studies on the prediction of terminators, most of them only focused on one kind 53 of them. In 2004, Wan XF, Xu D et al. proposed a prediction method for Rho-54 independent terminators with an accuracy of 92.25%. In 2005, Michiel J. L. de Hoon 55 et al. studied the sequence of Rho-independent terminators in B. subtilis[2], and the 56 final prediction accuracy was 94%. In 2011, Magali Naville et al. conducted a research 57 on Rho-dependent transcriptional terminators[3]. They used two published algorithms, 58 Erpin and RNA motif, to predict terminators. The specificity and sensitivity of the final 59 results were 95.3% and 87.8...