Centre Bretagne-Normandie

Abstract Background: Improving feed efficiency (FE) is an important breeding goal due to its economic and environmental significance for farm animal production. The phenotypic value is obtained by measuring individual feed consumption and average daily gain during a test period, which is costly and time-consuming. The identification of reliable predictors of FE may be a relevant strategy to reduce phenotyping efforts. Results: Gene expression data in the whole blood from three originally separated experiments were combined and analyzed by machine learning algorithms to propose predictive molecular biomarkers of FE traits in growing pigs. The original datasets included pure Large White pigs (females and barrows) of two lines divergently selected for residual feed intake (RFI), a measure of net FE, and in which individual feed conversion ratio (FCR) and blood microarray data were available. Merging the three datasets allowed considering FCR values (Mean = 2.85; Min = 1.92; Max = 5.00) for a total of n = 148 pigs with a large range of body weight (15 kg to 115 kg) and different test period duration (2 weeks to 9 weeks). Random forest (RF) and gradient tree boosting (GTB) were applied on the whole blood transcripts (26,322 annotated molecular probes) to identify the most important variables for binary classification on RFI groups and for a quantitative prediction of FCR, respectively. Samples have been partitioned between learning (n = 74) and validation sets (n = 74). Iterative steps in variable selection led to identify about three hundreds (328 to 391) molecular probes as important predictors for RFI or FCR and participating to various biological pathways. With the GTB algorithm, simpler models combining 34 expressed unique genes to classify pigs on RFI (100% of success), and 25 expressed unique genes to predict FCR values (R² = 0.80, RMSE = 8%) were proposed. Accuracy performance of RF models were slightly lower in classification and markedly lower in regression.Conclusion: From small subsets of genes expressed in the whole blood, it is possible to predict feed efficiency traits in growing pigs. This offers good perspectives for microchip design in selection or precision farming applications.