Computational Prediction of Viral Entry Dynamics: Spike Protein-Receptor Binding Affinity and Escape Mutations

Introduction

Viral entry into host cells is a critical step in the infection cycle, mediated by surface glycoproteins that recognize and bind specific host receptors [1, 2]. For coronaviruses, the spike (S) protein facilitates membrane fusion after binding to angiotensin-converting enzyme 2 (ACE2) or other entry receptors [3, 4]. The receptor-binding domain (RBD) within the S1 subunit undergoes conformational rearrangements to engage the receptor, a process governed by electrostatic and van der Waals interactions [5, 6]. Mutations in the RBD can substantially alter binding affinity, modulate host tropism, and enable immune evasion [7, 8]. Computational methods have become indispensable for predicting these changes before experimental validation, allowing rapid risk assessment for emerging zoonotic pathogens [9, 10].

In veterinary medicine, understanding host range and spillover potential is vital for surveillance and vaccine design [11, 12]. Many coronaviruses circulate in animal reservoirs such as bats, cats, and livestock, and computational pipelines help evaluate cross-species transmission risk [6, 11]. This article reviews the biophysical and algorithmic foundations for predicting spike protein-receptor binding affinity and escape mutations, focusing on molecular dynamics simulations, free energy perturbation, and machine learning models. We discuss structural bioinformatics tools, case studies on SARS-CoV-2 variants and bat coronaviruses, and implications for global genomic surveillance.

Molecular Dynamics Simulations and Free Energy Calculations

Molecular dynamics (MD) simulations provide atomic-level insight into spike protein-receptor interactions [13, 14]. By solving Newton's equations of motion for a solvated protein system, MD captures conformational fluctuations and transient contacts critical for binding [15, 16]. All-atom MD simulations of the RBD-ACE2 complex typically span tens to hundreds of nanoseconds, using force fields such as CHARMM or AMBER [17, 18]. These simulations reveal key stabilizing interactions, including hydrogen bonds, salt bridges, and hydrophobic contacts [19, 20]. For example, the RBD residue K417 forms a salt bridge with ACE2 D30, a contact that is disrupted in some variants [21, 22].

Free energy perturbation (FEP) and thermodynamic integration compute binding free energy differences between wild-type and mutant complexes [23, 24]. In FEP, the Hamiltonian is gradually transformed from one state to another, and the work required is averaged over multiple trajectories [25, 26]. These methods can predict changes in binding affinity (ΔΔG) with root-mean-square errors of approximately 1–2 kcal/mol [27, 28]. A key application is the prospective evaluation of emergent mutations in the RBD, such as N439K, Y453F, and N501Y, which were shown to enhance ACE2 affinity before experimental confirmation [29, 30].

Enhanced sampling techniques, including replica exchange MD and metadynamics, overcome kinetic barriers and map the free energy landscape of receptor engagement [31, 32]. Such approaches have identified intermediate states of the spike protein that may be targeted by therapeutics or antibodies [33, 34]. Constant pH MD has also been applied to model protonation states of histidine residues in the heptad repeat 1 (HR1) domain, affecting fusion kinetics [16].

Table 1 summarizes common computational methods for binding affinity prediction.

Table 1. Computational methods for predicting spike protein-receptor binding affinity.

Machine Learning and Structural Bioinformatics Pipelines

Machine learning (ML) models have accelerated the prediction of mutation effects on binding affinity and immune escape [3, 7]. Deep neural networks trained on large mutational scanning datasets can classify mutations as affinity-enhancing or escape-associated [17, 20]. Features such as evolutionary conservation, residue depth, and local structural environment are combined with contact maps derived from MD simulations [18, 21]. For example, attribute-guided latent space exploration (AGLSE) coupled with classical MD enabled the design of peptide inhibitors with high predicted affinity for the RBD [7].

Structural bioinformatics pipelines integrate multiple tools. A typical workflow begins with homology modeling or AlphaFold2 to generate three-dimensional structures of spike proteins from sequence data [11, 35]. The quality of the model is assessed via Ramachandran plots and MolProbity scores. Next, protein-protein docking tools, such as ClusPro or HADDOCK, predict the orientation of the RBD-ACE2 complex [13, 29]. Interface residues are then identified, and single-point mutations are introduced computationally using Rosetta or FoldX [28, 30]. MD simulations refine the complex, and MM/GBSA or FEP calculates binding energies.

graph TD
    A[Viral Spike Sequence], > B[Structure Prediction: AlphaFold2 / Homology Modeling]
    B, > C[Receptor Structure: ACE2 (host)]
    C, > D[Protein-Protein Docking: Orientation Prediction]
    D, > E[Interface Residue Identification]
    E, > F[In Silico Mutagenesis: Rosetta / FoldX]
    F, > G[Molecular Dynamics: GROMACS / NAMD]
    G, > H[Binding Free Energy: FEP / MM/GBSA]
    H, > I[Predicted ΔΔG & Escape Score]
    I, > J[Experimental Validation / Surveillance]
    J, > K[Updated Surveillance Guidance]
    K, > B

Ensemble learning methods that combine predictions from multiple force fields and sampling protocols improve robustness [5, 8]. For bat-derived coronaviruses, such as HKU5-CoV-2, structure-based virtual screening of FDA-approved antivirals was combined with MD to identify potential entry inhibitors targeting the S1 subunit [6].

Case Studies: SARS-CoV-2 Variants and Bat Coronaviruses

SARS-CoV-2 Variants

The emergence of SARS-CoV-2 variants with altered transmissibility and immune escape has been extensively studied using computational methods [28, 30]. The D614G mutation in the spike protein was predicted to increase ACE2 binding affinity via enhanced RBD opening, a finding confirmed by cryo-electron microscopy [33, 34]. Later variants, including Alpha (N501Y), Delta (L452R, T478K), and Omicron (multiple RBD mutations), were evaluated prospectively with MD and FEP [4, 20]. Omicron's BA.1 sublineage carries 15 RBD mutations; computational predictions indicated higher ACE2 affinity coupled with reduced antibody neutralization [15, 17]. Deep mutational scanning data were used to train ML classifiers that correctly identified escape mutations from class 1, 2, and 3 neutralizing antibodies [7, 20].

Bat Coronaviruses

Bat coronaviruses represent a major reservoir for future zoonotic spillover [6, 8]. Computational prediction of their spike protein binding to host ACE2 orthologs is crucial for risk assessment [11, 12]. For instance, the spike protein of HKU5-CoV-2, a bat merbecovirus, was modeled and docked against ACE2 from various mammals, including pigs, cattle, and rodents [6]. MD simulations showed that certain RBD residues (e.g., F486 in the loop region) are critical for binding, and mutations at those positions could enhance affinity for human ACE2 [29, 30]. A structure-based approach using hierarchical clustering of ACE2 species variants identified vertebrates with high predicted susceptibility to SARS-CoV-2, guiding surveillance efforts [11]. Similarly, Ma and Gong modeled ACE2 from frequently contacted animals (cats, dogs, horses) and found that feline ACE2 binds the SARS-CoV-2 RBD with high affinity, consistent with experimental infection data [12].

Implications for Vaccine Design and Global Genomic Surveillance

Computational prediction of binding affinity and escape mutations directly informs vaccine design. By identifying conserved epitopes that are resistant to mutation, immunogens can be engineered to elicit broad neutralizing responses [8, 23]. For example, the HR1 domain of the spike protein is conserved across coronaviruses, and computationally designed peptides targeting HR1 have shown broad inhibitory activity in vitro [4, 16]. Allosteric inhibitors that stabilize the RBD in its down conformation (inaccessible to ACE2) have been discovered through virtual screening and MD [19].

In global genomic surveillance, computational pipelines can prioritize newly detected spike mutations for experimental characterization. The framework shown in the Mermaid diagram allows near-real-time assessment of emerging variants. Repositories such as GISAID provide sequence data that feed into automated structure prediction and free energy calculations [3, 20]. For veterinary applications, similar pipelines can be established for influenza A virus hemagglutinin or African swine fever virus p72 to predict receptor binding changes [11, 12].

Conclusion

Computational methods have matured to the point where prediction of spike protein-receptor binding affinity and escape mutations is both rapid and reasonably accurate. Molecular dynamics simulations, free energy perturbation, and machine learning models each contribute unique strengths, and their integration into structural bioinformatics pipelines enables proactive risk assessment. The veterinary field benefits directly from these approaches by evaluating cross-species transmission potential and designing effective interventions. Continued improvements in force fields, sampling algorithms, and deep learning architectures will further enhance predictive power, supporting global preparedness for emerging zoonotic threats.

References

[1] Chahir R, Redouane S, Galan J, et al. Computational discovery of SARS-CoV-2 viral entry inhibitory peptides from Androctonus mauretanicus scorpion venom: molecular docking and molecular dynamics simulations targeting the spike protein. Front Bioinform. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41938340/

[2] Lee S, Yoon SJ, Lim J, et al. PoMA-10: a dual-action antiviral disrupting SARS-CoV-2 Spike-ACE2 interaction and protecting lung tissue. Front Pharmacol. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41908829/

[3] Gul Z, Shahid SA, Abid OU, et al. Discovery of thiadiazole-based small-molecule inhibitors of SARS-CoV-2 spike-ACE2 interaction through integrated computational prediction and experimental validation. Sci Rep. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41484143/

[4] Sattar A, Jahromi BS, Ghahi FSJ, et al. Computational screening and molecular modeling of probiotic-derived peptides targeting the conserved HR1 domain of SARS-CoV-2 spike protein. Sci Rep. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/41461805/

[5] Ullah A, Waqas M, Ullah S, et al. Boswellic acid derived molecules as SARS-CoV-2 spike protein inhibitors: A comprehensive virtual screening, triplicate molecular dynamic simulation and biochemical validation. Curr Med Chem. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/41084244/

[6] Dubey A, Kumar M, Tufail A. Inhibiting viral entry of bat-derived coronavirus HKU5-CoV-2: Targeting spike protein S1 subunit with FDA-approved antivirals-A structural dynamics and energetics study. Bioorg Chem. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40865231/

[7] Ullah F, Xiao A, Ullah S, et al. Synergizing Attribute-Guided Latent Space Exploration (AGLSE) with Classical Molecular Simulations to Design Potent Pep-Magnet Peptide Inhibitors to Abrogate SARS-CoV-2 Host Cell Entry. Viruses. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40573419/

[8] Reilly CB, Moore J, Lightbown S, et al. Broad-spectrum coronavirus inhibitors discovered by modeling viral fusion dynamics. Front Mol Biosci. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40443526/

[9] Neander L, Hannemann C, Netz RR, et al. Quantitative Prediction of Protein-Polyelectrolyte Binding Thermodynamics: Adsorption of Heparin-Analog Polysulfates to the SARS-CoV-2 Spike Protein RBD. JACS Au. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/39886596/

[10] Huang J, Jin Y, Wu R, et al. Identification of apigenin as a multi-target inhibitor against SARS-CoV-2 by computational exploration. FASEB J. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/39718442/

[11] Kaushik R, Kumar N, Zhang KYJ, et al. A novel structure-based approach for identification of vertebrate susceptibility to SARS-CoV-2: Implications for future surveillance programmes. Environ Res. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35460633/

[12] Ma C, Gong C. ACE2 models of frequently contacted animals provide clues of their SARS-CoV-2 S protein affinity and viral susceptibility. J Med Virol. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/33755203/ *** Disclaimer: This article is for educational and informational purposes only. It is not intended to substitute for professional veterinary advice, diagnosis, treatment, or regulatory guidance. Always consult a licensed veterinarian or qualified specialist regarding animal health, disease diagnosis, and therapeutic decisions.

[13] Bekono BD, Onguéné PA, Simoben CV, et al. Computational discovery of dual potential inhibitors of SARS-CoV-2 spike/ACE2 and Mpro: 3D-pharmacophore, docking-based virtual screening, quantum mechanics and molecular dynamics. Eur Biophys J. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/38907013/

[14] Ramachandran B, Nadeem A, Mohanprasanth A, et al. Prediction of deleterious non-synonymous SNPs of TMPRSS2 protein combined with Molecular Dynamics Simulations and free energy analysis to identify the potential peptide substrates against SARS-CoV-2. J Biomol Struct Dyn. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/38592189/

[15] Ovchynnykova O, Booth JD, Cocroft TM, et al. In silico Study on Natural Chemical Compounds from Citric Essential Oils as Potential Inhibitors of an Omicron (BA.1) SARS-CoV-2 Mutants' Spike Glycoprotein. Curr Comput Aided Drug Des. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/38178668/

[16] Yánez Arcos DL, Thirumuruganandham SP. Structural and pKa Estimation of the Amphipathic HR1 in SARS-CoV-2: Insights from Constant pH MD, Linear vs. Nonlinear Normal Mode Analysis. Int J Mol Sci. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/38003380/

[17] Güner E, Özkan Ö, Yalcin-Ozkat G, et al. Determination of Novel SARS-CoV-2 Inhibitors by Combination of Machine Learning and Molecular Modeling Methods. Med Chem. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/37957860/

[18] Grosche VR, Souza LPF, Ferreira GM, et al. Mannose-Binding Lectins as Potent Antivirals against SARS-CoV-2. Viruses. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/37766292/

[19] Li T, Yan Z, Zhou W, et al. Discovery of a Potential Allosteric Site in the SARS-CoV-2 Spike Protein and Targeting Allosteric Inhibitor to Stabilize the RBD Down State using a Computational Approach. Curr Comput Aided Drug Des. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/37493168/

[20] Abhinand CS, Prabhakaran AA, Krishnamurthy A, et al. SARS-CoV-2 variants infectivity prediction and therapeutic peptide design using computational approaches. J Biomol Struct Dyn. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/36572420/

[21] Shanmugam A, Venkattappan A, Gromiha MM. Structure based Drug Designing Approaches in SARS-CoV-2 Spike Inhibitor Design. Curr Top Med Chem. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/36330617/

[22] Siddiqui S, Upadhyay S, Ahmad R, et al. Interaction of Bioactive Compounds of Moringa oleifera Leaves with SARS-CoV-2 Proteins to Combat COVID-19 Pathogenesis: a Phytochemical and In Silico Analysis. Appl Biochem Biotechnol. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35838886/

[23] Squeglia F, Romano M, Esposito L, et al. Structure-Based Development of SARS-CoV-2 Spike Interactors. Int J Mol Sci. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35628409/

[24] Alexpandi R, Gendrot M, Abirami G, et al. Repurposing of Doxycycline to Hinder the Viral Replication of SARS-CoV-2: From in silico to in vitro Validation. Front Microbiol. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35602049/

[25] Oyedara OO, Agbedahunsi JM, Adeyemi FM, et al. Computational screening of phytochemicals from three medicinal plants as inhibitors of transmembrane protease serine 2 implicated in SARS-CoV-2 infection. Phytomed Plus. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/35403085/

[26] Fujimoto KJ, Hobbs DCF, Umeda M, et al. In Silico Analysis and Synthesis of Nafamostat Derivatives and Evaluation of Their Anti-SARS-CoV-2 Activity. Viruses. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35215982/

[27] Lewis DSM, Ho J, Wills S, et al. Aloin isoforms (A and B) selectively inhibits proteolytic and deubiquitinating activity of papain like protease (PLpro) of SARS-CoV-2 in vitro. Sci Rep. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35140265/

[28] Celik I, Khan A, Dwivany FM, et al. Computational prediction of the effect of mutations in the receptor-binding domain on the interaction between SARS-CoV-2 and human ACE2. Mol Divers. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35138508/

[29] González-Paz L, Alvarado MJ, Hurtado-León ML, et al. Comparative study of SARS-CoV-2 infection in different cell types: Biophysical-computational approach to the role of potential receptors. Comput Biol Med. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35077937/

[30] Khan A, Hussain S, Ahmad S, et al. Computational modelling of potentially emerging SARS-CoV-2 spike protein RBDs mutations with higher binding affinity towards ACE2: A structural modelling study. Comput Biol Med. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/34979405/

[31] Al-Shuhaib MBS, Hashim HO, Al-Shuhaib JMB. Epicatechin is a promising novel inhibitor of SARS-CoV-2 entry by disrupting interactions between angiotensin-converting enzyme type 2 and the viral receptor binding domain: A computational/simulation study. Comput Biol Med. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/34942397/

[32] Grishin AM, Dolgova NV, Landreth S, et al. Disulfide Bonds Play a Critical Role in the Structure and Function of the Receptor-binding Domain of the SARS-CoV-2 Spike Antigen. J Mol Biol. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/34780781/

[33] Kunkel G, Madani M, White SJ, et al. Modeling coronavirus spike protein dynamics: implications for immunogenicity and immune escape. Biophys J. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34767789/

[34] Basu S, Chakravarty D, Bhattacharyya D, et al. Plausible blockers of Spike RBD in SARS-CoV2-molecular design and underlying interaction dynamics from high-level structural descriptors. J Mol Model. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34057647/

[35] Daood U, Gopinath D, Pichika MR, et al. Molecular Dynamic Simulation Search for Possible Amphiphilic Drug Discovery for Covid-19. Molecules. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/33921378/