Sequences of antibody discovered as therapeutic option for COVID-19 are required to be analyzed before subjecting the sequence for further analysis. Since all variable domains fold into a series of beta strands joined by loops in a very similar 3D shape, the complementarity-determining regions (CDRs) are six of these loops at the top, where these regions develop loops that extend from the surface of the antibody, will result in direct contact with the antigen (Davies and Chacko, 1993). Numbering each residue according to a conventional approach is very helpful for sequence comparisons and engineering due to the continuity of the antibody structural similarity. Precise identification and characterization of these antibody regions are crucial in development and modification of antibodies (Patel et al., 2023). These annotated CDRs establish a significant degree of variation in antibody structure (Wong et al., 2019). Hence, it is critical for recognizing CDR to ensure its binding to a specific antigenic molecule before posing modifications to the antibody.

Numbering schemes with different approaches and set of applications have been developed to standardize the annotation of CDRs. An early yet widespread approach for annotating CDRs is the Kabat numbering scheme, which detects hypervariable regions and relies on the antibody sequences alignment (Kabat, 1991). The 3D structure of the antibodies is the foundation of Chothia numbering scheme (Chothia and Lesk, 1987) which emphasizes the structural locations of CDRs and the protected framework areas that sustain them. An enhanced version of the original Chothia scheme, the Martin scheme, introduces more structural insights and improves the numbering to cover a greater number of spots (Martin and Thornton, 1996), however, it has not been widely utilized. The well-established and comprehensive IMGT numbering scheme, annotates immunoglobulin and T cell receptors (Lefranc et al., 2015). It offers a standardized framework for comparing different species by ensuring consistency across species and antibody types by defining CDRs using both sequence and structural data.

Immunogenicity of the antibody sequences is also predicted to assess the immunogenic response of the therapeutic antibody which ensures safety and effectiveness. Immunogenicity prediction analysis helps in determining whether the antibody sequences exhibit low immunogenicity by identifying significant epitopes and ensuring that they fall below thresholds associated with strong immune activation. These antibodies can enhance their feasibility and reduce detrimental immune responses in various patient populations (Harris and Cohen, 2024).

ANARCI (Dunbar and Deane, 2015), an online tool that offers to annotate variable domains of antibodies from various species, enabling precise identification of CDRs and their alignment for immunogenicity analysis, is widely used in several SARS-CoV-2 studies (Wang et al., 2022; Xu et al., 2021; Zhou et al., 2023). Antibody region-specific alignment (AbRSA) (Li et al., 2019), is also a platform to perform sequence analysis by delimiting the CDRs and antibody numbering for numerous antibodies targeting viral particles (Dănăilă and Buiu, 2022; Dzimianski et al., 2023; Singh et al., 2023).

The successive unfolding process of protein folding transforms the protein sequences of the SARS-CoV-2 binding antibodies, which are mostly composed of a linear sequence of amino acids, into a functional three-dimensional antibody structure (Poluri et al., 2021). The arrangement of the amino acids determines its basic structure. From this linear arrangement, localized folding results in the formation of secondary structures including alpha helices and beta sheets, which are fuelled by hydrogen bonds between adjacent amino acids. The intricate three-dimensional tertiary structure is its repercussions of the continuous folding of the secondary structure together with the inclusion of loops and turns of the antibody (Rehman et al., 2022).

Protein folding analysis provides many useful insights about the interaction of the antibody especially through identifying the structure of CDR loop formations, but this multifaceted process requires expensive and specialized equipment, making it a challenging task before computational tools are being employed (Benjin and Ling, 2020; Brito and Archer, 2020). But as time passes, using in silico technology, where protein modeling has allowed for generally reliable predictions to be made (Srivastava et al., 2018). The goal of protein modeling is to make use of a range of computer methods to analyze amino acid sequences to predict the three-dimensional (3D) structure of the antibody sequences. Protein modeling provides distinctive approaches for predicting protein structures through a variety of tools that has been included on Table 2, which uses the protein sequences as an input (Agnihotry et al., 2022).

AlphaFold2 (Cheung et al., 2023; Ruff and Pappu, 2021; Skolnick et al., 2021) produces remarkably accurate 3D structure predictions using a neural network architecture that has been trained on a large database of structural and protein sequence data. This tool is utilized in various SARS-CoV-2-related studies that explore the binding behavior of its structural proteins (Ali and Caetano-Anollés, 2024; Jiao et al., 2023; Raisinghani et al., 2024). There are also several studies on the structural analysis of antibodies that prove the modeling capability of AlphaFold2 for antibody sequences (Du and Huang, 2023; Yin et al., 2022). SWISS-MODEL (Schwede et al., 2003) utilizes a homology-modeling approach that is performed iteratively until a satisfactory model structure is obtained. 3D structures of SARS-CoV-2 antibodies (Schepens et al., 2021; Beshnova et al., 2022) were successfully determined through this tool. MODELLER (Šali and Blundell, 1993) is a 3D modeling standalone tool, used to predict the 3D structure of SARS-CoV-2 antibodies (Mercurio et al., 2021; Yang et al., 2021) and restore missing residues in its structure (Martí et al., 2022; Giron et al., 2020). RoseTTAFold is one of the modeling tools that uses neural network-based techniques, incorporating connection between sequences, atomic coordinates, residue-residue orientations, and distances. This tool has been used in several studies on SARS-CoV-2 antibody discovery (Ford et al., 2022; Jing et al., 2024; Lubin et al., 2021). ABodyBuilder (Leem et al., 2016) is an antibody modeling software that incorporates multiple tools, including ABangle (Dunbar et al., 2013) and FREAD (Choi and Deane, 2010). Since this tool is specialized for antibody modeling, numerous studies employed ABodyBuilder to model the variable region of antibodies (Das et al., 2022; Das et al., 2023; Rouet et al., 2023) which also includes bispecific antibody (Ojha et al., 2022).

The specificity of a novel or pre-existing antibody of SARS-CoV-2 can be accessed through validation in silico using computational tools. The binding properties of an antibody are primarily determined by the sequence and structure of CDRs through molecular docking. Molecular docking is performed using the analyzed and modeled 3D antibody structures to study the interaction by predicting the preferred orientation, affinity, and interaction of an antibody-antigen complex by analyzing intermolecular interactions (Koçer and Çelik, 2024).

Molecular docking is a process that anticipates atomic-level molecular interactions (Agu et al., 2023). Molecular docking can be performed with various types of biological molecules which include small molecules such as drugs, metabolites, ligands, inhibitors, ions (Jarad et al., 2023; Noreen et al., 2023), and complex molecules that comprise DNA, RNA, proteins, peptides, carbohydrates, nucleosides (Aziz et al., 2023; Madku et al., 2023; Weng et al., 2020). According to research by (Shahmirzaie et al., 2020), molecular docking has proven its capability of being a pioneering analysis to validate biological model interaction by providing binding site information. In the process of validation of antibody binding, molecular docking helps in predicting the preferred orientation of an antibody to the targeted antigen when these molecules are bound to each other to form a stable complex (Gaudreault et al., 2023).

Binding of an antibody exhibits both rigid and flexible properties which is essential for efficient antigen recognition and immune response (Fernández-Quintero et al., 2020). Electrostatic interactions and complementary structures lead to a relatively rigid and specific binding between the paratope and epitope where the rigidity ensures high-affinity binding and specificity (Zeng et al., 2023). On the other hand, the antibody also exhibits flexibility that facilitates conformational changes in the antigen and antibody. Flexibility allows the antibody to bind to a wide range of epitopes and identify antigens with various conformations by allowing it to accommodate variations in the antigen structure (Kilambi and Gray, 2017). An induced-fit mechanism takes place in binding conditions, where the conformational changes between the antigen and antibody are made upon binding to enhance their interactions. The flexibility of an antibody allows it to adapt to the structural alterations in the antigen and improves binding affinity (Bekker et al., 2020). In general, an antibody requires dynamic equilibrium between rigid and flexible phases upon its binding to the antigen.

RosettaDock is a docking approach that optimizes both rigid-body orientations and side-chain conformation (Lyskov and Gray, 2008). RosettaDock is used to perform docking of nanobodies against SARS-CoV-2 receptor-binding domain (RBD) (Yang et al., 2021), monoclonal antibodies against rare antigenic site of SARS-CoV-2 spike protein (Suryadevara et al., 2024) and a specific antibody against SARS-CoV-2 spike protein to improvise the binding affinity (Neamtu et al., 2023). ZDOCK uses Fast Fourier Transform (FFT) to yield less clash penalty in docking (Chen et al., 2003). Several studies employed ZDOCK to study the SARS-CoV-2 antibody-antigen interaction (Khan et al., 2020; Nath et al., 2021). HawkDock is an unique docking tool with integration of the ATTRACT docking algorithm and the MM/GBSA free energy that allows determination of antibody-antigen binding precisely (Weng et al., 2019). Docking is performed through this tool with nanobodies and therapeutic antibodies for interaction analysis (Shah and Woo, 2022; Yang et al., 2024). ClusPro is a widely used docking tool that has benchmarked against alternative docking tools in Critical Assessment of Predicted Interactions (CAPRI) studies (Kozakov et al., 2017). This tool employed to study the binding properties of SARS-CoV-2 spike protein RBD with nanobodies (Shang et al., 2024) and SARS-CoV-2 spike protein with monoclonal antibodies (Nath et al., 2021). High Ambiguity Driven Docking Approach (HADDOCK) harnesses biochemical and biophysical interaction data, including mutagenesis or chemical shift perturbation data from NMR titration experiments to obtain near-native results. Binding prediction of the antibodies discovered with the targeted site on SARS-CoV-2 is performed in several studies using this tool (Ford et al., 2022; Ford et al., 2023).

The developability of antibody models discovered using in silico approach for COVID-19 will be studied and validated as they can aligned with the real-time experimentally produced therapeutic antibodies. Molecular dynamics (MD) simulations offer a dynamic and comprehensive understanding of biomolecular behavior at the atomic level, and have developed to be an essential tool in the study of computational biophysics (Lemm et al., 2021). In the field of antibody design, MD simulations have shown to be very helpful as a reliable means of testing in silico designs, bridging the gap between computational predictions and experimental findings by providing insights into the structures.

MD simulations operate based on the basic principles of classical mechanics, which make use of Newton’s equations of motion to predict the motions of individual atoms in a molecular system (Shukla and Tripathi, 2020). MD simulations accurately depict the interactions between atoms, including the flexibility of bonds, angle bending, and non-bonded interactions such as van der Waals forces and electrostatics, by applying a force field, a mathematical model that defines the potential energy of the system (Badar et al., 2022). The force field selection is essential to the precision and dependability of MD simulations since it significantly impacts the simulation outcomes. Numerous force fields with unique strengths and applications have been developed over time. CHARMM force field is one of the most common and extensible force fields in computational chemistry which operates exceptionally well to simulate lipids, proteins, and nucleic acids (Brooks et al., 2009). AMBER force field is particularly utilized for proteins and nucleic acids (Wang et al., 2004). The goal of AMBER is to supply precise parameter sets for biomolecular systems. The temporary conformational state of antibody binding is not always visible in static crystal structures but only can be revealed by MD simulations. Accurate parameterization of these forcefields in MD simulation play pivotal roles in comprehending the principles underlying antibody binding and refining antibody architectures to enhance their affinity and specificity for target antigens (Shaw et al., 2010).

GROMACS, an open-source software package designed for molecular dynamics simulations of biochemical molecules including proteins, acts as an in silico to study the behavior of antibody and antibody-antigen complexes at the atomic level (Berendsen et al., 1995; Van Der Spoel et al., 2005). The stability of various SARS-CoV-2 antibody-antigen complexes, including complexes involving the SARS-CoV-2 S protein and bispecific antibodies, as well as the SARS-CoV-2 S protein trimer with monoclonal antibodies, was assessed by measuring the root-mean-square fluctuation (RMSF) of the complexes to quantify dynamic stability (Ford et al., 2022; Ford et al., 2023).