Scholarlink

Vaccine development involves a complex integration of different computational analysis, experimental testing, and precise clinical trials. In silico screening accelerates antigen identification, which is critical for vaccine target selection. This study employs Vaxi-DL, an online-based deep learning platform, to predict protein fasta sequences that could serve as vaccine antigens for bacteria, protozoa, fungi, and viruses. Datasets of positive and negative sequences were compiled from known vaccine candidates and a database called Protegen, with physicochemical and biological properties, was assessed by using different bioinformatics tools. The data were partitioned into training, validation, and testing for each pathogen. Fully Connected Layers (FCLs) were used in model construction, with hyperparameters optimized for performance. Vaxi-DL’s parameters—including AUC, sensitivity, precision, specificity, and accuracy—were benchmarked against VaxiJen and Vaxign-ML, showing superior predictive ability. Testing on 219 potential vaccine candidates (PVCs) across 37 pathogens resulted in 175 accurate predictions, demonstrating 93% sensitivity. For addressing Chagas Vaxi-DL, along with Vaxelan, among the 17 proteins offered to us by one of our U.S. collaborators, a set of KMP 11 and other proteins having high S.i value have been suggested as vaccine candidates. Vaxelan, helps in the antigen assessment options using different immunoinformatics based tools like PsortB or WoLF PSORT, or even by using sub-cellular localization, epitopes, and virulence factors prediction for vaccine designing. While Vaxi-DL is a machine learning based online tool with vaxelan integrated in it.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning-Based Prediction of Vaccine Candidates from the Selected Proteins of T. Cruzi Using Vaxi-DL: Integrating Immunoinformatics Tools

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yuktika Malhotra,
  • Deepika Yadav,
  • Jerry John,
  • Kamal Rawal

摘要

Vaccine development involves a complex integration of different computational analysis, experimental testing, and precise clinical trials. In silico screening accelerates antigen identification, which is critical for vaccine target selection. This study employs Vaxi-DL, an online-based deep learning platform, to predict protein fasta sequences that could serve as vaccine antigens for bacteria, protozoa, fungi, and viruses. Datasets of positive and negative sequences were compiled from known vaccine candidates and a database called Protegen, with physicochemical and biological properties, was assessed by using different bioinformatics tools. The data were partitioned into training, validation, and testing for each pathogen. Fully Connected Layers (FCLs) were used in model construction, with hyperparameters optimized for performance. Vaxi-DL’s parameters—including AUC, sensitivity, precision, specificity, and accuracy—were benchmarked against VaxiJen and Vaxign-ML, showing superior predictive ability. Testing on 219 potential vaccine candidates (PVCs) across 37 pathogens resulted in 175 accurate predictions, demonstrating 93% sensitivity. For addressing Chagas Vaxi-DL, along with Vaxelan, among the 17 proteins offered to us by one of our U.S. collaborators, a set of KMP 11 and other proteins having high S.i value have been suggested as vaccine candidates. Vaxelan, helps in the antigen assessment options using different immunoinformatics based tools like PsortB or WoLF PSORT, or even by using sub-cellular localization, epitopes, and virulence factors prediction for vaccine designing. While Vaxi-DL is a machine learning based online tool with vaxelan integrated in it.