RT Journal Article
SR Electronic
T1 NeoMUST: an accurate and efficient multi-task learning model for neoantigen presentation
JF Life Science Alliance
JO Life Sci. Alliance
FD Life Science Alliance LLC
SP e202302255
DO 10.26508/lsa.202302255
VO 7
IS 4
A1 Ma, Wang
A1 Zhang, Jiawei
A1 Yao, Hui
YR 2024
UL https://www.life-science-alliance.org/content/7/4/e202302255.abstract
AB Accurate identification of neoantigens is important for advancing cancer immunotherapies. This study introduces Neoantigen MUlti-taSk Tower (NeoMUST), a model employing multi-task learning to effectively capture task-specific information across related tasks. Our results show that NeoMUST rivals existing algorithms in predicting the presentation of neoantigens via MHC-I molecules, while demonstrating a significantly shorter training time for enhanced computational efficiency. The use of multi-task learning enables NeoMUST to leverage shared knowledge and task dependencies, leading to improved performance metrics and a significant reduction in the training time. NeoMUST, implemented in Python, is freely accessible at the GitHub repository. Our model will facilitate neoantigen prediction and empower the development of effective cancer immunotherapeutic approaches.