[en] The current context leads to more and more efficient and rational animal productions. The objective of the “Genotype plus Environment” project (G plus E) is to support novel phenotyping approaches to provide large scale phenotypes for a genomic study and contributing to the sustainability of dairy cow production systems. In this framework, 3 European farms (AFBI-UK, UCD-IRL, AU-DK) collected observations (weight, body condition score, uterine health, residual feed intake, lameness,…) and samples (milk, blood, liver, feed,…) on 135 dairy cows, from calving until day 49. Those data constitute a substantial database which permits to link those phenotypes of interest to potential biomarkers, and especially the mid infrared (MIR) spectra of milk. Predicting phenotypes of interest from milk MIR spectra could be very interesting to detect specific status of cows in a cost effective, rapid and routine process, allowing the acquisition of data at large scale. Classification models have been developed from milk MIR spectra. For example a model built on 60 observations permits to distinguish animals with or without lameness with a good predicted classification of 68 and 71% respectively. Otherwise Regression models have been performed to predict molecules of interest from milk MIR spectra. Some of them can be used with a threshold (eg. level of milk NAGase which is associated to an inflammation status) some others present potential to be predicted quantitatively (eg. IGF1 which is linked to uterine health). This database therefore allows developing tools to predict new health indicators from milk MIR spectra that can be easily implemented at large scale. Those observations will be validated through new data collected with the same protocol from 3 other European farms.
