DPCMHC: efficient prediction of MHC-peptide binding affinity by deep learning based on dual-padding convolution
Lu, Y.; Mi, L.; Zhang, S.
Show abstract
Predicting peptide binding affinities to major histocompatibility complex class I (MHCI) is a crucial challenge in immunological bioinformatics and is essential for identifying neoantigens in personalized cancer vaccines. Current deep learning methods often struggle, especially with 10-mer and 11-mer peptides. To address this, we developed DPCMHC, an advanced deep learning model featuring an embedding module, a dual-padding convolutional module, a BiLSTM module, and an output module. This design enhances the models understanding of amino acid sequences and their lengths. DPCMHC effectively captures the information on the beginning and end of amino acids, as well as the diverse sizes of adjacent amino acids. Using concatenation, the model extracts continuous sequence information. We rigorously evaluated DPCMHC on three benchmark datasets, demonstrating its superior or comparable performance to existing state-of-the-art methods. Our validation results confirm that DPCMHC is a robust and efficient tool for predicting MHC-peptide binding affinities.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- BERTrand - peptide:TCR binding prediction using Bidirectional Encoder Representations from Transformers augmented with random TCR pairing 96%
- Machine learning optimization of peptides for presentation by class II MHCs 95%
- EPIC-TRACE: predicting TCR binding to unseen epitopes using attention and contextualized embeddings 95%
Similar papers in this journal
- Personalized deep learning of individual immunopeptidomes to identify neoantigens for cancer vaccines 94%
- The pitfalls of negative data bias for the T-cell epitope specificity challenge 93%
- Improving protein function prediction with synthetic feature samples created by generative adversarial networks 93%
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.