PromoterLCNN: A Light CNN-Based Promoter Prediction and Classification Model

Abstract

Promoter identification is a fundamental step in understanding bacterial gene regulation mechanisms. However, accurate and fast classification of bacterial promoters continues to be challenging. New methods based on deep convolutional networks have been applied to identify and classify bacterial promoters recognized by sigma (σ) factors and RNA polymerase subunits which increase affinity to specific DNA sequences to modulate transcription and respond to nutritional or environmental changes. This work presents a new multiclass promoter prediction model by using convolutional neural networks (CNNs), denoted as PromoterLCNN, which classifies Escherichia coli promoters into subclasses σ70, σ24, σ32, σ38, σ28, and σ54. We present a light, fast, and simple two-stage multiclass CNN architecture for promoter identification and classification. Training and testing were performed on a benchmark dataset, part of RegulonDB. Comparative performance of PromoterLCNN against other CNN-based classifiers using four parameters (Acc, Sn, Sp, MCC) resulted in similar or better performance than those that commonly use cascade architecture, reducing time by approximately 30–90% for training, prediction, and hyperparameter optimization without compromising classification quality.

Sigma70Pred: A highly accurate method for predicting sigma70 promoter in Escherichia coli K-12 strains

“…We have also considered the existing methods for performance comparison on the independent dataset, which were trained and evaluated on different datasets such as MULTiPly ( Zhang et al, 2019 ), iPromoter-2L ( Liu et al, 2018 ), and, iPromoter-2L2.0 ( Liu and Li, 2019 ). Moreover, to compare the efficiency of our generated model with deep-learning based classifiers, we have compared the performance with methods like iPromoter-BnCNN ( Amin et al, 2020 ), pcPromoter-CNN ( Shujaat et al, 2020 ), and PromoterLCNN ( Hernandez et al, 2022 ). We have calculated the different performance measures for all the working sigma promoter predictors.…”

“…On the other hand, computational approaches are reliable and fast with equivalent accuracy. Although, several methods have been developed in the past for the prediction of sigma promoters in the DNA sequences based on machine-learning ( Lin and Li, 2011 ; Song, 2012 ; He et al, 2018 ; Liu et al, 2018 ; Lai et al, 2019 ; Liu and Li, 2019 ; Zhang et al, 2019 ) and deep-learning approaches ( Amin et al, 2020 ; Shujaat et al, 2020 ; Hernandez et al, 2022 ), but the accurate identification of the sigma promoters remained a strenuous task due to the inter-and intra-class similarities and variations in the different sigma-specific promoter sequences ( Zhang et al, 2019 ). It has been seen in the past that promoter sequences often differ at one or more locations from the consensus sequences ( Mrozek et al, 2014 , 2016 ), which makes the task of prediction of sigma70 promoters more difficult as sigma70 factor specific promoters are responsible for the transcription of most of the genes in prokaryotic genome.…”

“…There are different methods which are specific to the classification of sigma70 promoters ( Lin and Li, 2011 ; Song, 2012 ; He et al, 2018 ; Lai et al, 2019 ; Rahman et al, 2019a , b ) whereas others are developed for the identification and classification of different sigma promoters such as sigma24, sigma28, sigma32, sigma38, sigma54, and sigma70 ( Liu et al, 2018 ; Liu and Li, 2019 ; Zhang et al, 2019 ; Amin et al, 2020 ; Shujaat et al, 2020 ; Hernandez et al, 2022 ). In this study, we have also developed a bioinformatic-ware to classify the sigma70 promoters using only sequence information.…”

“…Therefore, number of methods used the middle regions of long coding sequences of E. coli. K-12 genome to create the negative/non-promoter dataset ( Shujaat et al, 2020 ; Hernandez et al, 2022 ), whereas, other methods used both the coding and non-coding regions to extract non-promoter sequences ( Lin and Li, 2011 ; He et al, 2018 ; Lai et al, 2019 ; Liu and Li, 2019 ; Rahman et al, 2019a , b ; Amin et al, 2020 ). In the benchmark dataset used in this study, half of the negative samples or non-promoter sequences were extracted from the coding and rest half were obtained from convergent intergenic spacers (non-coding regions).…”

“… Shujaat et al (2020) introduced pcPromoter-CNN which convert the nucleotide sequence information into one-hot encoding vectors and feed them to convolutional neural network (CNN)-based classifier to predict and determine the sigma promoter classes. Recently, a new method based on the light CNN named as PromoterLCNN was proposed by Hernandez et al (2022) which also used one-hot encoding representation of nucleotide sequences to predict the sigma promoters using the sequencing information. The correct prediction of sigma70 promoters in the DNA sequences is still a difficult challenge due to the intraclass variation in terms of consensus sequence as sigma70 factor is responsible for the transcription of the most of the regulatory genes.…”