jbs-product-logo-main 1331898601

Study of Population Structure and Genetic Prediction of Buffalo from Different Provinces of Iran using Machine Learning Method Pages 48-59

Zahra Azizi, Hossein Moradi Shahrbabak, Seyed Abbas Rafat, Mohammad Moradi Shahrbabak and Jalil Shodja

DOI: https://doi.org/10.6000/1927-520X.2019.08.03.7

Published: 05 June 2020


Abstract: Considering breeding livestock programs to milk production and type traits based on existence two different ecotypes of Iranian’s buffalo, a study carried out to investigate the population structure of Iranian buffalo and validate its classification accuracy according to different ecotypes from Iran (Azerbaijan and North) using data SNP chip 90K by means Support vector Machine (SVM), Random Forest (RF) and Discriminant Analysis Principal Component (DAPC) methods. A total of 258 buffalo were sampled and genotyped. The results of admixture, multidimensional scaling (MDS), and DAPC showed a close relationship between the animals of different provinces. Two ecotypes indicated higher accuracy of 96% that the Area Under Curve (AUC) confirmed the obtained result of the SVM approach while the DAPC and RF approach demonstrated lower accuracy of 88% and 80 %, respectively. SVM method proved high accuracy compared with DAPC and RF methods and assigned animals to their herds with more accuracy. According to these results, buffaloes distributed in two different ecotypes are one breed, and therefore the same breeding program should be used in the future. The water buffalo ecotype of the northern provinces of Iran and Azerbaijan seem to belong to the same population.

Keywords: Classification, Buffalo, Machine learning, SNP Chip data.


Submit to FacebookSubmit to TwitterSubmit to LinkedIn