Similarity Estimation Between DNA Sequences Based on Local Pattern Histograms of Binary Images

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Alignment-free DNA sequence comparison is crucial for analyzing large genomic datasets.
• Graphical representations of DNA sequences are widely used for this purpose.
• Existing methods may face challenges with computational efficiency for very long sequences.

Purpose of the Study:

• To propose a novel, efficient method for DNA sequence feature extraction and similarity estimation.
• To utilize binary image representations and local bitmap pattern frequency histograms.
• To assess the method's applicability for phylogenetic analyses.

Main Methods:

• Representing DNA sequences as binary images.
• Extracting features using frequency histograms of local bitmap patterns.
• Evaluating sequence similarity with five different distance measures, focusing on histogram intersection and Manhattan distance.

Main Results:

• The proposed method achieves linear time complexity relative to DNA sequence length.
• This efficiency makes it suitable for comparing extensive sequences, including whole genomes.
• Histogram intersection and Manhattan distance were identified as the most effective measures for phylogenetic analysis.

Conclusions:

• The novel image-based feature extraction method offers a computationally efficient approach for DNA sequence comparison.
• The technique is scalable and practical for analyzing large-scale genomic data.
• Specific distance measures (histogram intersection, Manhattan distance) enhance its utility in phylogenetic studies.