Quantum mechanical electronic and geometric parameters for DNA k-mers as features for machine learning.

Area of Science:

• Genomics
• Computational Biology
• Biophysics

Background:

• Machine learning models in genomics often simplify DNA to basic nucleotide sequences (A, T, G, C).
• This simplification overlooks valuable scientific insights into nucleic acid structure and properties.
• There is a need for richer molecular features to improve genomic prediction models.

Purpose of the Study:

• To create a comprehensive database of quantum mechanical (QM) and geometric features for all 7-meric DNA permutations.
• To provide pre-computed molecular features for B, A, and Z DNA conformations.
• To facilitate the integration of detailed molecular information into DNA sequence analysis and modeling.

Main Methods:

• Utilized high-cost, time-consuming quantum mechanical (QM) methodologies.
• Calculated QM and geometric features for all permutations of 7-meric DNA.
• Generated a database of these pre-computed features for various DNA conformations (B, A, Z).

Main Results:

• Developed a comprehensive database containing QM and geometric features for 7-meric DNA sequences.
• Demonstrated the utility of these features by building a predictive model for A->C mutation rates.
• Enabled seamless association of novel molecular features to DNA sequences via k-meric window scanning.

Conclusions:

• The developed database offers a rich source of molecular information for DNA sequence analysis.
• Incorporating QM and geometric features significantly enhances the predictive power of genomic models.
• This approach advances machine learning applications in genomics by leveraging detailed molecular properties.