Introduction
# What is DLpMHCI?
DLpMHCI, short for Deep Learning peptide-MHC I, is an ensemble learning framework based software that predicts antigen
binding to HLA class I molecular. The performance of DLpMHCI is superior than other models because it incorporates the
residual network and the attention mechanism. DLpMHCI can reliably predict whether peptides with lengths between 8 and
10 mer might bind to 12,745 HLA class I molecules after being trained on the dataset of183,313 peptides from95HLA
class I alleles .
# Why is DLpMHCI?
- More Accurate Datasets
The training set for DLpMHCI is made up of very precise single allele mass spectrometry data rather than conventional binding affinity data. The single allele mass spectrometry data may precisely depict the peptide binding to a particular allele as well as the mechanism of peptide processing in vivo. It benefits from low false positive rates and high rates of accuracy.
- Biochemically Comprehensible Coding
To better reflect the biological features of the sequence and improve biological interpretability, DLpMHCI adopts amino acid physicochemical characteristics coding as opposed to the conventional BLOSUM coding.
- An Ensemble Learning Framework Based Model
In order to extract features, DLpMHCI incorporates self-attentional mechanisms and residual networks. It then classifies characteristics using a complete connection layer and determines if the output peptide is provided by HLA. We demonstrate that the integrated learning framework improves the model's prediction performance.
- Improved Prediction Accuracy
DLpMHCI performs better than other models in test sets, independent test sets, and antigen screening tests.
# Citation
# - name: OpenHarmony
# desc: 开放原子开源基金会
# link: https://docs.openharmony.cn/pages/000000/
# bgColor: '#f1f1f1'
# textColor: '#2A3344'
- name: OpenHarmonyOpenHarmonyOpenHarmonyOpenHarmonyOpenHarmony.
desc: xxxxxx,submitted
link: https://www.iqiyi.com/v_10331zk5kcg.html?vfm=2008_aldbd&fv=p_02_01
bgColor: '#f1f1f1'
textColor: '#2A3344'