A review of computational algorithms for CpG islands detection

Area of Science:

• Epigenetics and Genomics
• Bioinformatics

Background:

• CpG islands are key epigenetic regulatory regions involved in DNA methylation and gene promoter activity.
• Precise identification of CpG islands is essential for understanding biological functions but remains a challenge.
• Experimental and computational methods are employed for CpG island detection, with a focus on computational approaches to reduce cost and time.

Purpose of the Study:

• To review and compare the latest computational methods for detecting CpG-enriched regions (CpG islands).
• To provide insights into algorithms utilizing clustering, patterns, and physical-distance parameters for CpG island identification.
• To assist researchers in prioritizing and developing more accurate and efficient CpG detection algorithms.

Main Methods:

• Review of computational CpG detection methods, including those based on clustering, patterns, and physical-distance parameters.
• Comparative analysis of algorithms employing window-based, Hidden Markov Model, density, and distance/length-based approaches.
• Evaluation of methods applied to human and mammalian genomes for CpG island detection.

Main Results:

• Various computational tools exist for CpG island detection, differing in principles and parameters.
• Comparative analysis helps in selecting algorithms based on specific datasets and genomes for improved accuracy and sensitivity.
• Existing methods still face challenges, including the need for reduced false-positive rates.

Conclusions:

• Understanding the principles of different computational CpG detection tools is vital for accurate island identification.
• Prioritizing and developing efficient algorithms with lower false-positive rates is an ongoing need.
• This review facilitates informed choices in selecting and advancing computational strategies for CpG island detection.