A naukri.com initiative
Arxiv
3d
55
Image Credit: Arxiv
UniPTMs: The First Unified Multi-type PTM Site Prediction Model via Master-Slave Architecture-Based Multi-Stage Fusion Strategy and Hierarchical Contrastive Loss
- UniPTMs is the first unified framework for multi-type PTM prediction, addressing limitations of existing deep learning models.
- The framework utilizes a 'Master-Slave' dual-path collaborative architecture, incorporating Bidirectional Gated Cross-Attention module and Low-Dimensional Fusion Network.
- Additional components like Multi-scale Adaptive convolutional Pyramid and Bidirectional Hierarchical Gated Fusion Network enhance the feature integration capabilities of UniPTMs.
- UniPTMs showcases significant performance improvements over state-of-the-art models with the use of a novel Hierarchical Contrastive loss function and a lightweight variant named UniPTMs-mini.
Read Full Article
3 Likes
For uninterrupted reading, download the app