Estimating the entropy of DNA sequences

Area of Science:

• Bioinformatics
• Computational Biology
• Information Theory

Background:

• Shannon entropy is a standard measure for symbol sequence order, commonly applied to DNA.
• Estimating higher-order entropies (block entropies) is crucial for understanding symbol correlations in sequences.
• Existing methods face challenges with small observation numbers relative to possible outcomes.

Purpose of the Study:

• To present an assay for estimating higher-order entropies in DNA sequences with limited observational data.
• To reconstruct the underlying n-mer probability distribution using statistical principles.
• To compare the information-theoretic properties of DNA sequences with other data types.

Main Methods:

• Reconstruction of n-mer probability distributions using the theorem of asymptotic equi-distribution and the Maximum Entropy Principle.
• Application of constraints to ensure reconstructed distributions reflect real-world characteristics.
• Selection of the highest entropy solution among compatible distributions as the most probable one.
• Development and testing of an algorithm on DNA model sequences and the Epstein Barr virus genome.

Main Results:

• The developed algorithm accurately estimates higher-order entropies for DNA sequences.
• Comparison with texts, computer code, and music reveals unique properties of DNA sequences.
• DNA sequences exhibit a higher degree of freedom in symbol combinations compared to other information carriers.

Conclusions:

• The presented assay provides a robust method for analyzing complex DNA sequence information.
• The findings suggest that biological sequences may utilize a broader combinatorial space than artificial information systems.
• This research offers new insights into the information-theoretic nature of the genome.