DeepImmuno is Now Integrated into the Vecura Platform

What is DeepImmuno?

DeepImmuno is a powerful deep-learning toolkit specifically designed to predict the immunogenicity of peptide-MHC class I pairs and generate novel immunogenic peptides for T-cell vaccine development. It comprises two core components: DeepImmuno-CNN, which evaluates candidate peptides for their likelihood of triggering a CD8+ T-cell response, and DeepImmuno-GAN, which generates synthetic 10-mer peptides. This dual approach helps researchers bridge the gap between discriminative screening and de novo antigen discovery. It is especially useful for neoantigen prioritization in cancer immunotherapy and for bootstrapping vaccine candidate pools where limited data exists.

What can users do with DeepImmuno on Vecura?

With DeepImmuno on Vecura, users can:

• Predict the immunogenicity score of specific 9-mer or 10-mer peptide-HLA class I combinations.
• Generate 64 unique synthetic 10-mer immunogenic peptides tailored for HLA-A*0201.
• Prioritize high-potential neoantigen candidates derived from tumor mutation data for downstream validation.
• Streamline vaccine design workflows without needing to manage local deep-learning environments or complex model dependencies.

What the output means

The output provides a Bayesian posterior probability (an immunogenicity score between 0 and 1) for specific peptide-HLA pairs, where higher values indicate a stronger predicted potential to elicit a T-cell response. Additionally, the tool provides a list of synthetic 10-mer sequences generated by the GAN.

This output should be used to support scientific decision making. It does not replace experimental validation.

Why this matters

The effective identification of immunogenic peptides is a critical bottleneck in the development of precision immunotherapies and T-cell-based vaccines. By leveraging structural determinants like paratope embeddings and physicochemical amino acid properties, DeepImmuno enables researchers to sift through vast candidate pools, focusing limited laboratory resources on peptides with the highest likelihood of clinical success.

This integration simplifies the computational burden for researchers, allowing them to focus on experimental verification and therapeutic application rather than technical infrastructure.

• Developed by: Frank Li and colleagues (Cincinnati Children's Hospital Medical Center)
• Source: Official GitHub Repository
• Reference: Li, F., et al. (2021). DeepImmuno: deep-learning-based prediction of the immunogenicity of peptide-HLA class I pairs. Briefings in Bioinformatics. https://doi.org/10.1093/bib/bbab160