shengyongniu / seqtu

rSeqTU - a machine learning based R package for identification of bacterial transcriptional units

rSeqTU - a machine learning based R package for identification of bacterial transcriptional units

These R scripts will be wrapped up as an R package. Users will be able to install this package from Bioconductor.

Introduction

A transcriptional unit (TU) is composed of one or multiple consecutive genes on the same strand of a bacterial genome. The genes within a TU are transcribed into a single mRNA to respond to specific growth conditions, and the TU are regulated by one promoter. To delineate the transcriptional regulatory networks, it is a crucial first step to identify TUs within a bacterial genome accurately. To allow users to efficiently perform TU identification on their machine and provide a more accurate prediction, we develop an R package, named rSeqTU. rSeqTU R package can automatically select essential TU features through a random forest algorithm in a machine learning framework. Besides, rSeqTU performs nearly 98% accuracy in most of our testing TU cases, e.g., public RNA-Seq datasets of E. coli. Users will be able to install the rSeqTU package from Bioconductor and input their customized RNA-Seq dataset to conduct de-multiplexing, quality controlling, reads alignment, random-forest-based feature selection, prediction model training, and TU prediction. Moreover, rSeqTU presents results in graphical visualizations and interactive tables for customized downstream analysis. rSeqTU also output read count matrix of both genes and TUs for further differentially expression analysis.