ORF Discovery Using High-Resolution Ribosome Profiling Data in Arabidopsis

Identifying protein-coding genes is essential as proteins are usually the ultimate players in the cell. Recently, ribosome profiling (deep sequencing of ribosome footprints) has revealed that ribosomes occupy many regions in the transcriptome outside of the annotated protein-coding sequences, such as 5’UTRs and non-coding RNAs. To distinguish real translation events from contaminants in ribosome purification, we developed an efficient ribosome profiling protocol in Arabidopsis, which allows one to precisely infer codons being translated within ribosome footprints. Subsequently the footprint data are subjected to a modified Fourier Transform method to statistically test whether the periodicity of footprints of a given ORF reflects translocation patterns of actively translating ribosomes. With stringent criteria, ORFs from over 18,000 genes have been identified that have excellent agreement with the current annotation. In addition, several hundred upstream open reading frames (uORF) within 5’UTRs as well as a few dozen novel ORFs from transcripts annotated as non-coding have been identified with high coverage and stringent statistics. Our approach not only provides experimental support for predicted gene models but also discovery of previously unannotated ORFs. As the Arabidopsis genome is among the best annotated, we expect this approach will be even more powerful when applied to other species.