ORFhunteR: An accurate approach to the automatic identification and annotation of open reading frames in human mRNA molecules[Formula presented]

Abstract

The coding potential of RNA molecules can be estimated using algorithms that find open reading frames (ORFs). However, previously developed algorithms show limited performance. We developed a computational approach dedicated to the automatic identification of ORFs in a large set of human mRNA molecules. It is based on the vectorization of nucleotide sequences followed by classification using a random forest. The predictive model was validated on human mRNA molecules from the NCBI RefSeq and Ensembl databases and demonstrated almost 95% accuracy in detecting true ORFs. Our method is implemented into a powerful R/Bioconductor package ORFhunteR.