Detection of genomic regions associated malformations in newborn piglets: a machine-learning approach

Abstract

Background A significant proportion of perinatal losses in pigs occurs due to congenital malformations. The purpose of this study is the identification of genomic loci associated with fetal malformations in piglets. Methods The malformations were divided into two groups: associated with limb defects (piglet splay leg) and associated with other congenital anomalies found in newborn piglets. 148 Landrace and 170 Large White piglets were selected for the study. A genome-wide association study based on the gradient boosting machine algorithm was performed to identify markers associated with congenital anomalies and piglet splay leg. Results Forty-nine SNPs (23 SNPs in Landrace pigs and 26 SNPs in Large White) were associated with congenital anomalies, 22 of which were localized in genes. A total of 156 SNPs (28 SNPs in Landrace; 128 in Large White) were identified for piglet splay leg, of which 79 SNPs were localized in genes. We have demonstrated that the gradient boosting machine algorithm can identify SNPs and their combinations associated with significant selection indicators of studied malformations and productive characteristics. Data availability Genotyping and phenotyping data are available at http://www.compubioverne.group/data-and-software/.

Keywords

• Congenital malformations
• Machine learning
• GWAS
• Agriculture
• Pigs