Computational design of a personalized mRNA vaccine candidate using exome data from a single breast cancer case
Personalized vaccine design tested with public exome data
A study analyzed publicly available breast cancer case data from Pakistan, selected seven candidate neoantigens, and computationally designed an antigen incorporating them for a personalized mRNA vaccine. This is a proof of concept based on a single case; it does not demonstrate an immune response or therapeutic benefit from an actual vaccine.
The researchers cited rising breast cancer incidence and mortality among women in Pakistan as the background for research into personalized immunotherapy. They used public whole-exome sequencing (WES) data from a cohort in Khyber Pakhtunkhwa. The data are registered with the U.S. National Center for Biotechnology Information (NCBI) under BioProject PRJNA941166. The study focused on demonstrating a personalized vaccine design workflow by analyzing all sequencing runs linked to the single available case.
Seven candidates selected from more than 43,000 variant peptides
Somatic variant analysis reported 6,005 missense variants that change amino acids across genes including MUC3A and TTN. From more than 43,000 candidate variant peptides, the researchers prioritized seven neoantigens predicted to bind strongly to human leukocyte antigen (HLA) and not to be allergenic. The stated HLA-binding energy threshold was ΔG ≤ −13.0 kcal/mol. These are computational screening results; they do not mean that antigen presentation or an immune response was confirmed in the patient.
The selected epitopes—the antigen regions recognized by the immune system—were combined with a granulocyte-macrophage colony-stimulating factor (GM-CSF) adjuvant and helper epitopes. The designed multi-epitope antigen is 285 amino acids long.
The AlphaFold2 structure model and computational quality assessments yielded a ProSA Z-score of −7.14 and an ERRAT score of 96.59%. The researchers interpreted these results as supporting the stability of the designed structure. They also reported that the structure remained stable in a 500-nanosecond molecular dynamics simulation. Interactions with the innate immune receptors known as Toll-like receptors (TLRs) were also predicted; the binding free energies were −148.8 kcal/mol for TLR9 and −16.7 kcal/mol for TLR2. These computational results do not establish actual receptor activation or immune efficacy.
From predicted immune responses to preclinical validation
The C-IMMSIM immune simulation suggested a response skewed toward type 1 helper T cells (Th1). Its predicted outputs included cytotoxic T-lymphocyte induction, memory T-cell formation, and increased interferon gamma (IFN-γ). These are all simulation outputs and should be distinguished from immune responses measured in patients or animals.
The study’s contribution is a workflow that selects candidates from public variant data, assembles an antigen, and computationally assesses its structure and immune response. The authors interpreted it as a basis for further preclinical evaluation. However, the findings described in the provided abstract extend only to computational predictions for a single case. No results confirming actual antigen expression and presentation, T-cell responses, safety, or therapeutic benefit were presented. Whether the approach applies to other patients also remains to be tested.
Source: Article citation and abstract indexed in Europe PMC.
UNLABELLED: Breast cancer is a common cancer type that occurs among women in Pakistan, and the rising incidence and mortality rate underline the need to develop effective, patient-tailored immunotherapies. In this study, we implemented an end-to-end immunoinformatics workflow using publicly available whole-exome sequencing (WES) data deposited under NCBI BioProject PRJNA941166, a cohort-derived resource from the Khyber Pakhtunkhwa region; as a proof of concept, we analyzed all sequencing runs associated with the available case to demonstrate a personalized vaccine design workflow. Somatic variant analysis indicated a high mutational burden, including 6005 missense mutations in genes such as MUC3A and TTN. From > 43,000 candidate mutant peptides, we prioritized seven non-allergenic neoantigens with strong predicted HLA binding (ΔG ≤ - 13.0 kcal/mol). These epitopes were assembled into a 285-amino acid multi-epitope antigen incorporating a GM-CSF adjuvant and helper epitopes. AlphaFold2 modeling and in silico quality assessment supported construct stability (ProSA Z-score - 7.14; ERRAT 96.59%). Across 500 ns molecular dynamics simulations, the vaccine construct remained conformationally stable and showed favorable predicted interactions with innate immune receptors, with strong binding free energies for TLR9 (ΔG = - 148.8 kcal/mol) and TLR2 (ΔG = - 16.7 kcal/mol). Immune simulations using C-IMMSIM suggested a Th1-skewed response characterized by induction of cytotoxic T lymphocytes, memory T-cell formation, and elevated IFN-γ. Although limited to computational predictions and a single publicly available case, the predicted receptor engagement and immunogenicity provide a rationale for preclinical evaluation of this personalized mRNA vaccine design workflow in high-risk populations. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s40203-026-00631-6.
This study is a computational design example linking patient-derived public exome data to candidate vaccine antigens. Follow-up researchers could consider the selected neoantigens and multi-epitope construct for preclinical evaluation and test whether actual antigen expression and presentation and T-cell responses match the computational predictions. However, the provided abstract does not include candidate sequences or specific experimental conditions, so the full text and supplementary materials would need to be reviewed before reproducing and validating the work. Predictions from a single case cannot establish the efficacy, safety, or manufacturability of a patient-specific vaccine. The current data also do not establish development speed or cost in resource-limited settings, a reduction in breast cancer mortality, or a preventive effect.