359. From raw sensor and automated data to genetic evaluation and validation in the cloud

Abstract

Precision livestock farming tools, such as sensor and automation techniques, enable animal breeders to define novel traits. New challenges are data storage and handling and the development of genetic evaluation. Traditionally, the development of genetic evaluation for novel traits requires several steps, i.e. data curation, trait definition, variance component estimation (ASReml), genetic evaluation (MiXBLUP), and validation of the estimated breeding values (EBVs), requiring many iterations to optimize the genetic evaluations. We combined these steps in a cloud solution to make this entire process more efficient. An experiment was run with 1,782,373,113 daily milk yield records of 1,120,550 cows, all the way from data curation to trait definition and validation of the EBV. The resulting wall-time, i.e. elapsed actual time from start to finish ofthe entire process, was ~23 hours. The flexible cloud solution can be easily modified or adapted to develop novel traits.