Identifying compositionally homogeneous and nonhomogeneous domains within the human genome using a novel segmentation...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Mammalian genomes are thought to consist of long, compositionally homogeneous domains.
• Current segmentation algorithms, like D(JS) using Jensen-Shannon divergence, struggle with identifying domain boundaries and halting criteria.
• Existing methods exhibit intrinsic biases in domain detection.

Purpose of the Study:

• To address the limitations of current genome segmentation algorithms.
• To introduce IsoPlotter, a novel parameter-free algorithm for identifying compositionally homogeneous domains.
• To compare IsoPlotter's performance against existing methods using simulated and real genomic data.

Main Methods:

• Developed IsoPlotter, a parameter-free segmentation algorithm utilizing a dynamic halting criterion.
• Implemented and compared IsoPlotter with the D(JS) algorithm on simulated genomic sequences.
• Applied IsoPlotter to segment the human genome to analyze domain composition.

Main Results:

• IsoPlotter successfully identified both long and short compositionally homogeneous domains with low GC content dispersion.
• The D(JS) algorithm failed to detect short domains and sequences with low compositional dispersion.
• Analysis of the human genome revealed one-third nonhomogeneous domains and a mix of short and long homogeneous domains.

Conclusions:

• IsoPlotter offers an unbiased and effective approach for genome segmentation.
• The human genome comprises a significant proportion of nonhomogeneous regions and a diverse range of homogeneous domain lengths.
• IsoPlotter provides a more accurate method for understanding genome composition and structure.