Reverse Vaccinology and Deep Learning for identifying bacteria protective antigens BPAgs
- Utilized Reverse Vaccinology and designed an MLP method for identifying bacteria protective antigens BPAgs with biological and physiochemical features annotated using bioinformatics software such as psortB, SPAAN, SignalP, TMHMM, IEDB, and Propy.
- Enhanced model applicability against new emerging pathogens by implementing a leave-one-pathogen-out strategy and benchmarked on a curated independent dataset. Model achieved an AUC-ROC score of 0.94, prediction accuracy 95%, and weighted F1 score 0.95.
