Popcorn: prediction of short coding and noncoding genomic sequences in prokaryotes.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Identifying prokaryotic genes, especially short open reading frames (sORFs) and small proteins, presents a significant challenge.
• RNA sequencing (RNA-seq) reveals numerous small transcripts of unknown function, complicating the distinction between coding and noncoding sequences.

Purpose of the Study:

• To develop a novel machine learning method for accurately classifying prokaryotic DNA sequences as either coding or noncoding.
• To address the difficulty in determining the coding potential of small transcripts identified through RNA-seq.

Main Methods:

• Introduction of Popcorn (PrOkaryotic Prediction of Coding OR Noncoding), a machine learning approach designed for sequence classification.
• Validation of Popcorn's efficacy in distinguishing between coding and noncoding sequences.

Main Results:

• Popcorn demonstrates high effectiveness in differentiating coding sequences from noncoding sequences.
• The method successfully identifies coding short open reading frames (sORFs) and noncoding RNAs.

Conclusions:

• Popcorn provides a reliable computational solution for classifying prokaryotic sequences.
• This tool aids in the identification of novel small proteins and regulatory RNAs, advancing prokaryotic gene discovery.