Machine Learning-Guided Systematic Search of DNA Sequences for Sorting Carbon Nanotubes

Area of Science:

• Materials Science
• Nanotechnology
• Biotechnology

Background:

• Efficient DNA sequence screening is crucial for sequence-dependent applications.
• The vast number of DNA sequence combinations presents a significant challenge for systematic search.
• Previous empirical methods for DNA sequence search exhibit low efficiency and accuracy.

Purpose of the Study:

• To develop a systematic and efficient method for searching DNA sequences.
• To enable the sorting of single-chirality single-wall carbon nanotubes (SWCNTs) using DNA sequence recognition.
• To improve upon the efficiency and accuracy of existing DNA sequence search strategies.

Main Methods:

• Utilizing the sequence-dependent recognition between DNA and single-wall carbon nanotubes (SWCNTs).
• Integrating machine learning algorithms with experimental investigations for DNA sequence identification.
• Developing and applying pattern recognition from short DNA sequences (5-mer and 6-mer) for broader application.

Main Results:

• Achieved a significant increase in the number of resolving sequences from approximately 10^2 to 10^3.
• Improved the success rate of finding resolving sequences from around 10% to over 90%.
• Demonstrated the scalability of identified sequence patterns to longer DNA sequences.

Conclusions:

• The combined machine learning and experimental approach offers a highly efficient and accurate method for DNA sequence screening.
• This strategy effectively addresses the challenge of identifying specific DNA sequences for applications like SWCNT sorting.
• The findings pave the way for broader applications of DNA sequence recognition in materials science and nanotechnology.