The main goal of the CAUSEL project is to develop a procedure that improves the classic bovine selection. The project targets the two limiting factors of the classic selection: i) the SNP markers are not causal variants and ii) the statistical models only consider additive effects and neither dominance or epistasis.
- Idenfication of causal variants: More than 20 million polymorphisms or genetic variants are known today in bovines. A large fraction are considered “neutral” and do not have an effect on the phenotype. A minority of variants (around 1000 per phenotype) affect the functioning of genes and potentially one or more phenotypes. These “causal” variants can either modify the protein sequence of a gene or affect a gene switch and thus perturb the gene expression profile. We will use genomic methods based on next generation sequencing to systematically identify coding variants in BBB.
- Application of machine learning techniques to the classic selection: While current methods assume only additive effects, it is very likely that more complex interaction mechanisms such as dominance and epistasis play an important role. Another problem is that the number of variables (SNPs) is much greater than the number of available samples. Therefore, MLG will apply scalable and big data machine learning techniques to identify strategies that can improve the classic selection.