SenSeqNet: A Deep Learning Framework for Cellular Senescence Detection From Protein Sequences.

Area of Science:

• Biotechnology
• Computational Biology
• Gerontology

Background:

• Cellular senescence is a key driver of aging and age-related diseases.
• Accurate detection of senescence is crucial for understanding aging and developing therapies.
• Current methods for detecting senescence are often slow and difficult to scale.

Purpose of the Study:

• To develop a novel deep learning framework, SenSeqNet, for predicting cellular senescence directly from protein sequences.
• To improve the efficiency and scalability of senescence detection.

Main Methods:

• SenSeqNet integrates Evolutionary Scale Modeling (ESM-2) embeddings with a hybrid LSTM-CNN architecture.
• The model analyzes protein sequences to capture sequential and structural features relevant to senescence.

Main Results:

• SenSeqNet achieved 86.43% accuracy in independent testing, surpassing traditional machine learning and deep learning methods.
• High-confidence genes predicted by SenSeqNet were significantly enriched in senescence-associated pathways, confirming biological relevance.

Conclusions:

• SenSeqNet offers a robust, biologically informed tool for detecting cellular senescence.
• This AI framework can accelerate research into aging mechanisms and the development of anti-aging therapeutics.