Prediction of two novel overlapping ORFs in the genome of SARS-CoV-2

Abstract

Six candidate overlapping genes have been detected in SARS-CoV-2, yet current methods struggle to detect overlapping genes that recently originated. However, such genes might encode proteins beneficial to the virus, and provide a model system to understand gene birth. To complement existing detection methods, I first demonstrated that selection pressure to avoid stop codons in alternative reading frames is a driving force in the origin and retention of overlapping genes. I then built a detection method, CodScr, based on this selection pressure. Finally, I combined CodScr with methods that detect other properties of overlapping genes, such as a biased nucleotide and amino acid composition. I detected two novel ORFs (ORF-Sh and ORF-Mh), overlapping the spike and membrane genes respectively, which are under selection pressure and may be beneficial to SARS-CoV-2. ORF-Sh and ORF-Mh are present, as ORF uninterrupted by stop codons, in 100% and 95% of the SARS-CoV-2 genomes, respectively.

Computational methods for inferring location and genealogy of overlapping genes in virus genomes: approaches and applications

“…[ 22 • ]. In a subsequent study [ 25 ], SeqComp was applied to a dataset of about 4000 spurious overlapping genes, with the aim to calculate its specificity. A spurious overlapping gene consists of a protein-coding ORF which overlaps with another ORF purely by chance.…”

“…A spurious overlapping gene consists of a protein-coding ORF which overlaps with another ORF purely by chance. In analysis of this dataset, SeqComp showed a low specificity, as it classified as true negative only 43.1% of the spurious overlaps [ 25 ].…”

“…To overcome the low specificity of SeqComp, I extended prediction criteria of SeqComp to four and combined this method with a codon scrambling (CodScr) test to produce CodScr + SeqComp method [ 25 ]. In addition to the two criteria used by SeqComp [ 22 • ], this method included two other criteria.…”

“… High sensitivity and low specificity. CodScr + SeqComp [ 25 ] Detects overlapping genes based on their peculiar nucleotide and amino acid composition and a statistically significant bias in codon usage. Good sensitivity and specificity.…”

“… Type of overlap detected Does it provide a P-value in the results? Availability SeqComp [ 22 • ] Single nucleotide sequences Protein-protein coding No Not implemented CodScr + SeqComp [ 25 ] Single nucleotide sequences Protein-protein coding No Not implemented Codon test [ 26 ] Single nucleotide sequences Protein-protein coding Yes Script at http://github.com/TimSchlub/Frameshift GOPHIX [ 27 ] Single nucleotide sequences Protein-protein coding No Not implemented Synplot2 [ 28 , 37 ] Multiple homologous sequences Protein-protein coding or functional RNA element Yes Web site at http://www.firthlab.path.cam.ac.uk/virad.html FRESCo [ 29 ] Multiple homologous sequences Protein-protein coding or functional RNA element No Script at https://www.broadinstitute.org/fresco/fresco PhyloCSF [ 30 , 31 ] Multiple homologous sequences Protein-protein coding Yes Script at http://compbio.mit.edu/PhyloCSF cRegions [ 32 , …”