Immunogenicity Prediction with Dual Attention Enables Vaccine Target Selection

Authors: Song Li, Yang Tan, Song Ke, liang hong, Bingxin Zhou

ICLR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments demonstrate that VENUSVACCINE outperforms existing methods across a wide range of evaluation metrics.
Researcher Affiliation Collaboration 1 School of Physics and Astronomy, Shanghai Jiao Tong University. 2 School of Information Science and Engineering, East China University of Science and Technology. 3 Shanghai Matwings Technology Co., Ltd.
Pseudocode No The paper describes methods narratively and with equations, but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code Yes The implementation is at https://github.com/songleee/Venus Vaccine.
Open Datasets No All source data and the implementation to reproduce the results will be publicly available upon acceptance.
Dataset Splits Yes For each dataset, we randomly split the data in a 7 : 1 : 2 ratio and select the model with the highest accuracy on the 10% validation set for evaluation.
Hardware Specification Yes All implementations were done using Py Torch (version 1.7.0), and experiments were run on an NVIDIAr RTX 3090 GPU with 24GB VRAM, mounted on a Linux server.
Software Dependencies Yes All implementations were done using Py Torch (version 1.7.0)
Experiment Setup Yes The model is optimized using ADAMW (Loshchilov et al., 2017) with a learning rate of 0.0005 and a weight decay of 0.01. In the attention pooling layer, dropout was set to 0.1. The maximum training epoch was set to 50, with early stopping based on validation accuracy and patience of 5.