DNA physical properties outperform sequence compositional information in classifying nucleosome-enriched and...

Area of Science:

• Genomics
• Epigenetics
• Structural Biology

Background:

• Nucleosomes are fundamental units of eukaryotic chromatin, essential for epigenetic regulation of DNA replication, recombination, and transcription.
• Accurate identification of nucleosome positions is vital for understanding genome function.
• Previous models using DNA deformation energy showed promise in predicting nucleosome dyad positions and occupancy.

Purpose of the Study:

• To establish and verify a machine-learning model for predicting nucleosome occupancy.
• To further investigate the role of DNA physical descriptors in nucleosome positioning.
• To assess the predictive power of DNA physical properties compared to sequence composition.

Main Methods:

• Development of a machine-learning model for nucleosome occupancy prediction.
• Utilizing a set of DNA physical descriptors as input features.
• Comparison of the new model against sequence compositional information-based models.

Main Results:

• The developed machine-learning model outperformed existing sequence-based models.
• Distinct DNA physical descriptor features differentiate nucleosome-enriched and -depleted regions.
• DNA physical properties effectively characterize gene transcription start and termination sites.

Conclusions:

• DNA physical properties are stronger determinants of nucleosome positioning than sequence composition alone.
• Physical descriptors reveal distinct characteristics of nucleosome-bound and unbound genomic regions.
• DNA physical properties play a regulatory role in gene transcription, influencing key regulatory sites.