iPromoter-BnCNN: a novel branched CNN-based predictor for identifying and classifying sigma promoters

Abstract

<jats:title>Abstract</jats:title> <jats:sec> <jats:title>Motivation</jats:title> <jats:p>Promoter is a short region of DNA which is responsible for initiating transcription of specific genes. Development of computational tools for automatic identification of promoters is in high demand. According to the difference of functions, promoters can be of different types. Promoters may have both intra and inter class variation and similarity in terms of consensus sequences. Accurate classification of various types of sigma promoters still remains a challenge.</jats:p> </jats:sec> <jats:sec> <jats:title>Results</jats:title> <jats:p>We present iPromoter-BnCNN for identification and accurate classification of six types of promoters - σ24, σ28, σ32, σ38, σ54, σ70. It is a CNN based classifier which combines local features related to monomer nucleotide sequence, trimer nucleotide sequence, dimer structural properties and trimer structural properties through the use of parallel branching. We conducted experiments on a benchmark dataset and compared with six state-of-the-art tools to show our supremacy on 5-fold cross-validation. Moreover, we tested our classifier on an independent test dataset.</jats:p> </jats:sec> <jats:sec> <jats:title>Availability</jats:title> <jats:p>Our proposed tool iPromoter-BnCNN web server is freely available at http://103.109.52.8/iPromoter-BnCNN. The runnable source code can be found https://colab.research.google.com/drive/1yWWh7BXhsm8U4PODgPqlQRy23QGjF2DZ</jats:p> </jats:sec> <jats:sec> <jats:title>Supplementary information</jats:title> <jats:p>Supplementary data (benchmark dataset, independent test dataset, model files, structural property information, attention mechanism details and web server usage) are available at Bioinformatics. online.</jats:p> </jats:sec>

Promotech: A general tool for bacterial promoter recognition

“…To do this, we obtained Promotech’s predictions on a balanced data set with 2,860 experimentally validated E. coli promoters collected from RegulonDB [26]. This data set has been used to evaluate the performance of several E. coli promoter prediction tools [10, 12, 27]. The average 5-fold cross-validation MCC and accuracy reported on this data set [10, 12, 27] are in the range of [0.498, 0.763] and [0.748, 0.882], respectively.…”

“…This data set has been used to evaluate the performance of several E. coli promoter prediction tools [10, 12, 27]. The average 5-fold cross-validation MCC and accuracy reported on this data set [10, 12, 27] are in the range of [0.498, 0.763] and [0.748, 0.882], respectively. RF-HOT achieved on this data set a MCC of 0.54, accuracy of 0.77, AUPRC of 0.845 and AUROC of 0.84; while, RF-TETRA achieved a MCC of 0.47, accuracy of 0.734, AUPRC of 0.830 and AUROC of 0.808.…”

“…Di Salvo et al comparatively assessed G4PromFinder’s performance in terms of F1-score with that of bTSSfinder, PePPER [5] and PromPredict [6, 7], and found that G4PromFinder outperformed the other three tools in GC rich bacterial genomes. There are several recently published E. coli promoter prediction methods such as MULTiPly [8], SELECTOR [9], iPromoter-BnCNN [10], IBPP [11], iPromoter-2L [12], among others (Supplementary Table 1). Cassiano and Silva-Rocha [13] carried out a comparative assessment of bacterial promoter prediction tools for identifying E. coli σ 70 promoters.…”

pcPromoter-CNN: A CNN-Based Prediction and Classification of Promoters

“…The dataset used to train the model is imbalanced, so different techniques, such as Synthetic Minority Oversampling Technique (SMOTE), were used to overcome the problem; however, SMOTE can easily turn the model toward data overfitting. Being inspired from [21] we proposed the use of a cascading binary classifier. The problem identified in the proposed architecture of [21] was that it uses four different encoding schemes and a large number of convolutional filters, which eventually increases both computational cost and complexity.…”

“…MULTiPly first identifies the best combination of information features by using an F-score feature selection method; that step is followed by applying five binary classifiers to identify the promoter class. A model named iPromoter-BnCNN proposed by Amin et al utilized four parallel one-dimensional convolutional filters applied to the monomer nucleotide sequence, the trimer nucleotide sequence, and the structural belonging dimers and trimers of the DNA sequence [21]. The dense layer combined all of the extracted features and performed the classification task.…”