Multiple Lines of Evidence Support 199 SARS-CoV-2 Positively Selected Amino Acid Sites

Abstract

SARS-CoV-2 amino acid variants that contribute to an increased transmissibility or to host immune system escape are likely to increase in frequency due to positive selection and may be identified using different methods, such as codeML, FEL, FUBAR, and MEME. Nevertheless, when using different methods, the results do not always agree. The sampling scheme used in different studies may partially explain the differences that are found, but there is also the possibility that some of the identified positively selected amino acid sites are false positives. This is especially important in the context of very large-scale projects where hundreds of analyses have been performed for the same protein-coding gene. To account for these issues, in this work, we have identified positively selected amino acid sites in SARS-CoV-2 and 15 other coronavirus species, using both codeML and FUBAR, and compared the location of such sites in the different species. Moreover, we also compared our results to those that are available in the COV2Var database and the frequency of the 10 most frequent variants and predicted protein location to identify those sites that are supported by multiple lines of evidence. Amino acid changes observed at these sites should always be of concern. The information reported for SARS-CoV-2 can also be used to identify variants of concern in other coronaviruses.

Keywords: SARS-CoV-2; coronaviruses; positively selected amino acid sites.

Figure 1

Venn diagrams showing the overlap of the PSSs in the COV2Var database (COV2Var-int-5%) identified by each one of the methods used (FEL (blue), FUBAR (red), and MEME (green)), and those here identified (this work (yellow)), for the structural (S-E), non-structural (nsp1-16), and accessory (ORF3a-ORF10) proteins of SARS-CoV-2.

Figure 2

Pie chart showing the contribution (in percentage) of each species to PSSs identified in non-SARS-CoV-2 species.

Figure 3

PSS location on SARS-CoV-2 protein monomers. For each protein, from top to bottom: PSSs present in the COV2Var-int-5% list (in blue are PSSs, in gray are variable amino acid sites; no information is available for the E and NSP7 proteins), PSSs identified in this work (green—identified in this work and in the COV2Var-int-5% list; orange—identified in this work and in at least one other method in the COV2Var-int-5%; and red—only identified in this work; there are no PSSs for NSP2, NSP3, and NSP7), the top 10 variants (pink if it has a frequency over 5%, cyan otherwise; there is no information for ORF10), and regions identified as hot PSS regions in non-SARS-CoV-2 species (dark red). For NSP9, as well as for accessory proteins, there is no information for the latter.

Figure 4

(A) Location of PSSs (labeled in red) supported by more than one type of evidence in the SARS-CoV-2 S homotrimer protein structure (PDB accession number 7DF4). Each monomer is shown in a different color. The PDB accession numbers of the docking partners of the S protein are shown above the respective structure (PDB accession numbers 7S0D and 7DF4). (B) Consurf projection of the S trimer as in the Consurf Database and its respective color code [70]. The PDB accession number 7DF4 was used as the query.

Figure 5

(A) Location of PSSs (labeled in red) supported by more than one type of evidence in the SARS-CoV-2 RTC (PDB accession number 7EGQ). NSP7, NSP8, NSP9, NSP10, NSP12, NSP13, and NSP14 are labeled in yellow, violet, beige, brown, green, pink, and white, respectively. (B) Consurf projection of the replicase complex dimer as in the Consurf Database [70]. The PDB accession number 7EGQ was used as the query.