Abstract
Detecting high-order single-nucleotide polymorphism (SNP) interactions is of great importance for the discovery of pathogenic causes of human complex diseases. However, a considerable computing challenge exists in analyzing each SNP combination at a genome-wide scale. Swarm intelligence search (SIS) is an effective and efficient method for solving NP-hard problems and has been extensively researched for detecting high-order SNP interactions. In this review, we first analyze the strengths and limitations of existing methods such as exhaustive search using cluster computing and parallel computing, stochastic search and high-performance computing. Then, SIS algorithms for the detection of high-order SNP interactions are introduced in detail. The algorithms discussed are the genetic algorithm (GA), ant colony optimization (ACO), harmony search (HS), particle swarm optimization (PSO), differential evolution (DE), cuckoo search (CS), fish swarm (FS) and artificial bee colony (ABC). Finally, we discuss the characteristics and limitations of the involved methods and provide several suggestions for improving SIS algorithms to detect high-order SNP interactions.
Highlights
With the rapid progress of genome sequencing technology, the cost of genome-wide sequencing has been greatly reduced, and genome-wide data volumes have increased rapidly, making genome-wide association studies (GWAS) widely involved in detecting single-nucleotide polymorphisms (SNPs) associated with complex human diseases.During the past decade, thousands of associated SNPs have been successfully identified since the first GWAS for age-related macular degeneration was presented by Klein et al [1], with the main focus being on individual SNPs that are isolated based on their contribution to disease status
To compare the detection power of seven classical swarm intelligence search algorithms (GA, ant colony optimization (ACO), harmony search (HS), cuckoo search (CS), differential evolution (DE), particle swarm optimization (PSO), artificial bee colony (ABC) and artificial fish swarm (AFS)), we investigate them on 12 DME models
An Swarm intelligence search (SIS) algorithm conducts a global search through the power of the group, therein aiming to enhance the perception of individual searchers in the search space through communication and learning between individuals in the group
Summary
With the rapid progress of genome sequencing technology, the cost of genome-wide sequencing has been greatly reduced, and genome-wide data volumes have increased rapidly, making genome-wide association studies (GWAS) widely involved in detecting single-nucleotide polymorphisms (SNPs) associated with complex human diseases. Traditional stochastic search algorithms can discover some k-order SNP interactions, these algorithms are still insufficient for the detection of high-order SNP interactions when using posterior association probabilities of individual loci, especially for the detection of complex disease models with minimal or no marginal effects. SIS rapidly achieves an overall understanding in a complex environment through mutual communication and learning among individuals in the group and can be used to solve high-dimensional complex optimization problems. For the problem of detecting high-order SNP interactions from high-dimensional space, it is beneficial to employ the SIS method to accelerate the search process by discovering some candidate k-order SNP combinations that have an association with disease status. During the past ten years, SIS methods have attracted wide attention in the study of epistasis analysis
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
