The availability of gene expression data in vaccine trials has generated new opportunities for understanding and predicting the response to vaccine. It has led to the so-called Systems vaccinology. However, the analysis of such data is difficult because of the high dimensionality of the predictors (p) in regards of the number of available subjects (n). In this talk, I will present several methods inspired by these applications that we have developed in my team such as testing differential expression/abundance of genes, reducing dimensions while considering geneset structures and random forest with the Frechet metrics.