Table 2.

In silico prediction of the pathogenic mechanisms of SARS-CoV-2

Refs.	Viral/cellular subunit	Modelling method	Primary results
18	Receptor ACE2 + SARS-CoV-2 spike.	1) Structure alignment between SARS-CoV-2 and SARS-COV; 2) SARS-CoV-2 + ACE2 complex modeling via single point mutation from solved SARS-CoV +ACE2; 3) SARS-CoV-2 and SARS-CoV interaction with ACE2 hot spots (residues mainly involved in RBD binding) comparison; 4) comparison with animals ACE2.	The two spike proteins are quite similar and therefore the authors analyzed in detail the hot spots revealed for SARS-CoV; 2) two virus-binding hot spots identified: ACE2 Hot spot 31(salt bridge Lys31-Glu35) and Hot spot 353 (salt bridge Lys353-Asp38); 3) (a) SARS-CoV-2 Q493 (vs SARS-CoV N479) was compatible but had less of an affinity with hotspot 3, (b) SARS-CoV-2 Q501 (vs SARS-CoV T487) had less affinity with hotspot 353, but was still compatible, (c) SARS-CoV-2 L455 (vs SARS-CoV Y442) preferred interaction with Hotspot 31, (d) SARS-CoV-2 F586 (vs SARS-CoV L472) enhances interaction with Hotspot 31 (e) SARS-CoV-2 S494 (vs SARS-CoV D480) made the interaction weaker but compatible with the hotspot 353; 4) mouse or rat ACE2 contained a hotspot at the 353 position, which did not fit into the RBD of SARS-CoV-2 which instead recognized ACE2 from pigs, ferrets, cats, orangutans, monkeys, and humans.
19	Receptor ACE2 + SARS-CoV-2 spike.	1) Genomic analysis; 2) homology modeling for SARS-CoV-2 spike; 3) structural superimposition in SARS-CoV+ACE2 model; 4) binding free energy evaluation.	1) Low homology in RBD domain between SARS-CoV-2 and SARS-CoV, but high similarity (similar polarity); 2) From free energy binding the authors observed that even though it was weaker than SARS-CoV (-78.6 kcal mol-1), SARS-CoV-2 spike still showed an high affinity with ACE2 (-50.6 kcal mol-1); 3) Differences in aa sequences did not alter the protein conformation and maintained similar Van der Waals and electrostatic properties.
17	receptor CD26 + SARS-CoV-2 Spike.	1) Homology modeling for SARS-CoV-2 Spike (S1+S2) using the SARS-CoV as template (PDB: 6ACD) to obtain inactive configuration and using SARS-CoV+ACE2 complex structure as a template (PDB:6ACG) to obtain the active structure; 2) analysis of glycosylation sites via the server; 3) comparison (SARS-CoV-2 vs SARS-CoV) of the glycosylation shield structure obtained by server; 4) docking between the modeled trimeric spike and human CD26 (PDB: 4QZV).	1) SARS-CoV-2 Spike tetramer configuration in both active and inactive state; 2) Based on the Spike+ACE2 model the authors identified the 3C-like proteinase cleavage site + glycosylation sites; 3) from SARS-CoV and SARS-CoV-2 comparison: SARS-CoV-2 showed several unique glycolysation sites in addition to the conserved from SARS-CoV. This suggested a different shielding or camouflage pattern; 4) SARS-CoV-2 spike showed a large interaction surface with CD26 with many unique residues (different in SARS-CoV) involved in the interaction. SARS-CoV-2, compared to previous SARS-CoV viruses, interacts in a different way with the human receptors.
28	ORF1ab (Nsp2 and Nsp3).	Provides information on how quickly the virus could potentially increase its genetic variability, with implications for contagiousness and drugs via comparison of ORF1ab SARS structure: 1) The ORF1ab of 15 SARS-CoV-2 sequence alignment with five sequences of SARS virus and five sequences from Bat SARS‐like virus; 2) selective pressure analysis; 3) homology modeling for NSp2 Nsp3; 4) trans-membrane analysis.	Potential sites under positive selective pressure were found (501); transmembrane helices were predicted. At the residue 501, Bat SARS‐like coronavirus showed an apolar amino acid (threonine), while SARS and SARS-CoV-2 showed a polar amino acid, glycine and glutamine, respectively. Hypothesis: polarity, and potential to form H‐bonds may confer higher stability to the protein. 723 aa. SARS-CoV-2 sequence showed a Serine replacing for Glycine in Bat SARSlike and SARS coronaviruses. Hypothesis: substitution increases the local stiffness of the polypeptide chain due to a steric effect and to the ability of Serine side chain to form H‐bonds. At 1010 aa, SARS-CoV-2 has Prolin, the Bat SARS‐like coronavirus and SARS virus showed a polar and an apolar a. Hypothesis: steric bulge and stiffness of the Proline may cause local conformation perturbation compared with the proteins of the other two viruses.
27	Spike-ACE2.	Prediction of the interaction ACE2 RBD-spike. Evolutionary analysis and search for the possible virus reservoirs. Comparisons of the spike sequences between SARS‐CoV‐2 with SARS‐CoV, Bat SARS‐like CoV, and other coronaviruses. Analysis of the ACE2 structures and binding motif alignment:1) Alignment of spike protein sequences from different sources. Full-length and RBD sequences of spike protein from SARS‐CoV, bat, or pangolin SARS-like CoV and SARS‐CoV‐2 were aligned; 2) comparison of ACE2 enzyme among different species; 4) prediction of spike protein model and spike‐ACE2 binding model.	Bat SARS‐like CoV RaTG13, is an inner joint neighbor of SARS‐CoV‐2 (6.2% overall genome sequence identity); the full‐length sequence of S pangolin SARS‐like CoV and SARS‐CoV‐2 seem to be a little different but SARS‐CoV‐2 RBD sequence (329 to 521) like CoV, had a higher than 89% similarity with bat SARS‐like CoV; pangolin SARS‐like CoV has a higher probability to cross host barriers and infect humans; from infection-involved aa comparison, they identified pangolins, turtles as possible intermediate hosts.
26	ORF1ab.	1) The authors showed a typical workflow for homology modeling applied to the SARS-CoV-2 ORF1ab proteins; 2) aa sequence comparison. They performed sequence alignment investigations on 10 primary sequences of SARS-CoV-2 via BLAST, SWISS‐MODEL, and Clustal Omega.	1) According to BLAST analysis, the sequence identity of ORF1ab protein between SARS-CoV-2 and SARS‐CoV was over 90% with the query cover of about 100%; 2) the authors showed a typical workflow for homology modeling applied to the SARS-CoV-2 ORF1ab proteins.
29	Spike + ACE2 receptor.	This study explores the binding of the proteins encoded by different human ACE2 allelic variants with SARS‐CoV‐2 spike protein. 1) They selected the coding variants of ACE2 and predicted in silico the possible allelic variants; 2) they modelled the 17 obtained coding variants of ACE2 via homology modeling with native structure and superimposed the structures for comparison; 3) They superimposed the obtained models over the native ACE2‐spike protein complex; 4) inter‐residual interaction maps.	1) The authors selected the allelic structures involved in the ACE2-spike interaction (17 variants); 2) The protein architecture of ACE2 allelic variants was similar to the wild type but the spatial orientation of substituting residues varied notably; 3) ACE2 variants bind with spike protein identical to the native complex + intermolecular contacts between SARS‐CoV‐2 spike protein and ACE2 variants are comparable. Among the differences, rs73635825 (S19P) and rs143936283 (E329G) showed a low binding affinity, which may confer resistance against the virus infection.