Figure 2.

Overview of the FIT algorithm [19]. FIT consists of a compendium of 170 CSP of mouse and human transcriptomics data for 28 different diseases. First, a lasso regression is performed to fit parameters α and β of a linear model based on all genes, g, of all CSP, p, penalizing α values deviating from 0 and β values deviating from 1. The fitting process is repeated 100 times to obtain mean values and confidence intervals of the two parameters (see also main text). For a new mouse expression data set, the mean values of the parameters α and β are then used to predict human effects sizes Z^H for each gene therein. (Mouse clipart by Vincent Le Moign / CC BY 4.0.)