Related Experiment Video
Updated: Aug 11, 2026
Identification and Classification of Position-specific GABAA Receptor Subunit Missense Variants for Their Role In Hippocampal Pyramidal Neurons
Published on: June 6, 2025
A Deep Learning Model for Prediction of Unknown Gene Functionality: Gene Bio-BERT
Srinivas Kudipudi1, Vyshnavi Durga Chirumamilla1, Pavani Ippili1
1Department of Computer Science and Engineering, Department of Artificial Intelligence and Data Science, Siddhartha Academy of Higher Education Deemed to Be University, Vijayawada, Andhra Pradesh, India.
A new Gene Bio-BERT framework automates gene function prediction using deep learning, achieving 94.5% accuracy. This approach enhances biomedical data analysis by effectively predicting functions for both known and unknown genes.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- The Human Genome Project mapped the human genome, but predicting gene function remains challenging.
- Existing deep learning models like Bio-BERT and ProtBERT struggle with direct gene function prediction without further training.
- Automated gene function prediction is crucial for advancing biomedical data analysis.
Purpose of the Study:
- To propose a novel Gene Bio-BERT based framework for automated gene function prediction.
- To leverage deep learning methods for enhanced accuracy in predicting gene functionality.
- To overcome limitations of existing models in predicting gene function from biomedical data.
Main Methods:
- A 3-module framework: Data collection/preprocessing, Gene Bio-BERT model training, and feature aggregation/prediction.
- Utilized Entrez API from NCBI for human gene data retrieval.
- Employed a Gene Bio-BERT transformer encoder with an attention-based feature fusion layer for training.
Main Results:
- Achieved exceptional accuracy of 94.5% and an F1 score of 0.87 in predicting gene function.
- Demonstrated robust performance on unannotated genes, maintaining a similarity score of 0.84.
- The framework effectively learns contextual embeddings and aggregates features for accurate predictions.
Conclusions:
- The proposed Gene Bio-BERT framework significantly improves automated gene function prediction.
- Deep learning, specifically this framework, offers a powerful solution for analyzing complex genomic data.
- This method provides a reliable tool for identifying functions of unknown genes, advancing genomic research.
