نوع مقاله : مقاله پژوهشی
نویسندگان
1 استاد، دانشکده کشاورزی و منابع طبیعی دانشگاه تهران
2 گروه علوم دامی ، دانشگاه تهران
3 گروه علوم دامی، دانشگاه تهران
4 گروه علوم دامی، دانشگاه بوعلی همدان
5 گروه علوم دامی، دانشگاه تربیت مدرس
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Introduction
Growth traits are quantitative traits controlled by a large number of genes with small effects, each of which has an additive effect on the phenotype. Today, machine learning methods have been used in animal science not only in GWAS but also in other important topics such as image processing, genomic evaluation, and prediction of important traits. Machine learning methods and their algorithms such as deep learning, random forest, support vector machine, and boosting have been introduced into GWAS topics. The advantages of machine learning methods are their high potential and efficiency, especially for large-volume data and estimation of non-additive effects such as dominance and epistasis, as well as the investigation of complex relationships between variables (such as interactions between markers). Therefore, the aim of this study is to use new technologies, such as genome-wide association studies (GWAS) along with artificial intelligence-based methods such as gradient boosting, to identify mutations in loci associated with growth traits in Lori-Bakhtiari sheep.
Materials and Methods
For the present study, blood samples were collected from 132 Lori-Bakhtiari sheep. Genomic DNA was extracted from whole blood using the Cinagene kit (Cinagene Company, Iran). The samples were genotyped using Illumina 50K Ovine SNPChip, which included 51,135 SNP markers. The gbm package in R software was used to implement the GB method. All markers were ranked based on their importance after applying the gradient boosting method based on the best combination of parameters. After identifying these markers, the genes reported within ± 500 kb of the top 50 SNPs were identified using the NCBI and Ensembl databases from the sheep reference genome Oar_v4.0. To explore the biological and functional processes of the identified genes, ontology studies were performed using the DAVID online database. Finally, gene network analyses were performed using Cytoscape software.
Results and Discussion
Genetic parameters and birth weight breeding values were estimated in this study. After quality control, a total of 44,796 SNP markers from 122 sheep remained for final analysis. Manhattan plots were drawn for birth weight and genome-wide SNP values using relative influence. The distribution of ranked SNPs (from most important to least important) was shown based on GB analysis. Our results showed that 2,509 SNPs (5/6%) had positive effects and 42,287 SNPs (94/4%) had neutral effects. The results of the investigation of genes reported within or adjacent to the 50 significant SNP markers identified in the present study led to the identification of 32 candidate genes. Gene ontology analysis of the 50 significant SNP markers and 32 candidate genes identified showed significant associations between the identified ontology pathways and biological processes related to weight gain. Among the top 10 SNPs and the genes identified around them using gradient boosting, there are 4 important genes including KANK1, ZNF277, DOCK4, and APOD, which are involved in the biological processes of body weight traits.
Conclusion
In this study, The top 10 SNPs were identified and introduced as potential essential SNPs for analysis. 32 candidate genes for birth weight, some of which are introduced for the first time. The results of the present study can provide valuable information for future research to identify genes, precisely determine candidate genomic regions for important economic traits such as birth weight, and better understand biological mechanisms using gradient boosting algorithm in Iranian sheep breeds.
کلیدواژهها [English]
)