GENN: a GEneral Neural Network for learning tabulated data with examples from protein structure prediction.

Area of Science:

• Computational biology
• Machine learning
• Bioinformatics

Background:

• Accurate prediction of biological data is crucial for research.
• Existing methods may struggle with large datasets or specific data structures.

Purpose of the Study:

• Introduce a General Neural Network (GENN) for versatile data trend learning and prediction.
• Highlight GENN's efficiency with large datasets and windowed input/output.

Main Methods:

• Developed a two-layered neural network architecture.
• Implemented options for separate input-output pairs and window-based data structures.
• Utilized GENN for protein-specific prediction tasks.

Main Results:

• GENN demonstrated remarkable performance in predicting protein accessible surface area.
• Successfully applied GENN for scoring protein similarity to native structures.
• Achieved high accuracy in predicting protein disorder using GENN.

Conclusions:

• GENN offers a general, simple, and efficient tool for learning from and predicting with large datasets.
• The network's architecture effectively handles windowed data, proving valuable in bioinformatics.
• GENN shows significant promise for various computational biology and machine learning applications.