Immunoinformatics vaccine design against neurological disorders; Machine learning-based reverse vaccinology approaches

Backgrounds and Aims: Vaccines have been used to prevent infectious diseases. However, theiruse can be extended to confer protection against non-infectious diseases, such as neurologicaldisorders. Vaccines are designed in multiple forms, such as inactivated (killed), toxoid, and liveattenuated. Multiple approaches are employed in vaccine design. New advances in computationalmedicine/biomedical engineering and immunobiology have resulted in the emergence of novelvaccines through machine-learning based tools. We discuss new tools in the field of computationalvaccinology with emphasis on targeting neurological disorders.Methods: This is an expert perspective and narrative review that discusses frequently used bioinformaticsmethods in application of computational vaccine design in neurological disorders.We also review literature from PubMed and Scopus using the search terms Immunoinformatics,neurology, machine learning. No time limit has been considered for publication. mRNA and proteinvaccines are reviewed and common methodology are described for each stage of in silicovaccine design. We review most-frequently used machine learning-based tools used in reversevaccinology (RV).Results: The computational process of vaccine design starts with selection of biologically significantantigens in the pathogenesis of neurological disorders. For example, amyloid plaques area good selection for Alzheimer’s disease (AD). Also, viral proteins involved in nervous systeminfections may be targeted to design vaccines against CNS infections. Machine learning toolsare utilized to identify helper and cytotoxic T-cell as well as B-cell epitopes. Servers that carryout such processes include Immune Epitope Data Base (IEDB), and CTLPred (based on ANN,KNN). Linker peptides including ALL, AGGGA, SSL, and EAAAK are then used to connectthe epitopes. Linkers act as spacers that prevent the conformational clash of epitopes. A peptideadjuvant is sometimes inserted in the sequence to boost immunization. The final sequence shouldbe modelled to produce a tertiary structure of the vaccine. This can be performed using UCSFMODELLER, RaptorX server, GalaxyWeb. Next, the structure quality is verified by Ramachandranplot (based on torsion angles) and ProSA (based on NMR/X-ray structures’ Z-score). Thevaccine undergoes physicochemical and antigenic (VaxiJen server) properties analysis. Dockingwith immune receptors such as TLRs are also performed. To make mRNA vaccines, in silico toolsare utilized to design and predict the RNA sequence, as opposed to protein sequences which aremodelled in the process of protein vaccine construction. The interaction of Vaccine-receptor isconfirmed via molecular dynamics analyses (GROMACS, Schrodinger, amber, or others). Finally,the Immune inducing potential of vaccine and its booster dosages are simulated via the C-ImmSimserver. This server enables simulation of adaptive and innate immune response followingmultiple injections of vaccine with or without adjuvant.Conclusions: Using machine learning-based Bioinformatics tools, RV has produced candidates

کلیدواژه ها:

Neuroimmunology ، Immunoinformatics ، Machine learning ، Molecular dynamics simulation (MDS) ، Immune simulation ، Artificial neural network (ANN)