A Virtual Machine Platform for Non-Computer Professionals for Using Deep Learning to Classify Biological Sequences of...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Metagenomic data analysis requires robust biological sequence classification for tasks like species and gene function identification.
• The complexity of metagenomic data and the need for specialized algorithms pose challenges for biologists lacking computational expertise.
• Deep learning (DL) offers powerful solutions for classification, with recent advancements making DL frameworks more accessible.

Purpose of the Study:

• To provide a guideline for constructing an accessible deep learning framework for biological sequence classification.
• To empower biologists to perform sequence classification without extensive mathematical or programming knowledge.
• To facilitate the analysis of complex metagenomic data.

Main Methods:

• Development of an easy-to-use deep learning framework tailored for sequence classification.
• Optimization of all code within a virtual machine for direct user implementation.
• Guidance provided for constructing DL frameworks without requiring in-depth algorithmic knowledge.

Main Results:

• A user-friendly deep learning framework for biological sequence classification is presented.
• The framework allows biologists to build custom DL models for their specific needs.
• Optimized code enables direct application to user's own metagenomic data.

Conclusions:

• The developed deep learning framework democratizes advanced sequence classification for biologists.
• Researchers can now analyze metagenomic data more effectively, even without specialized computational skills.
• This approach accelerates discovery in fields relying on biological sequence analysis.