Design of Machine Learning Models for the Prediction of Transcription Factor Binding Regions in Bacterial DNA

Abstract

Transcription Factors (TFs) are proteins that regulate the expression of genes by binding to their promoter regions. There is great interest in understanding in which regions TFs will bind to the DNA sequence of an organism and the possible genetic implications that this entails. Occasionally, the sequence patterns (motifs) that a TF binds are not well defined. In this work, machine learning (ML) models were applied to TF binding data from ChIP-seq experiments. The objective was to detect patterns in TF binding regions that involved structural (DNAShapeR) and compositional (kmers) characteristics of the DNA sequence. After the application of random forest and Glmnet ML techniques with both internal and external validation, it was observed that two types of generated descriptors (HelT and tetramers) were significantly better than the others in terms of prediction, achieving values of more than 90%.

Computational Analysis of the Tripartite Interaction of Phasins (PhaP4 and 5)-Sigma Factor (σ24)-DNA of Azospirillum brasilense Sp7

“…Despite the above, there are well-characterized motifs for the binding of the σ 70 , σ 54 , σ 28 , and σ 32 factors to the −10 and −35 regions in Escherichia coli. However, due to the low conservation of the σ 24 and σ 28 factors, it has not been possible to find a well-characterized motif for DNA binding [18,20]. Also, it is necessary to consider the DNA sequences for interaction with the sigma factors [17].…”

Guidelines on the performance evaluation of motif recognition methods in bioinformatics

“…Nevertheless, many studies that present novel methods for motif detection repeatedly appear without adequate comparative evaluation. The main issues include comparison against no or only a single method, despite several comparable methods existing ( Alvarez-Gonzalez and Erill, 2021 ; Hammelman et al, 2022 ), the use of only one dataset, usually with unknown true positives ( Levitsky et al, 2022 ), and the use of uncommon statistical metrics ( Zhang et al, 2019 ). The last can be exemplified with a criterion of the correct prediction—if, within the top ten, there is a motif similar (not identical!)…”