Fig. 1
From: Dose–response prediction for in-vitro drug combination datasets: a probabilistic approach
The figure shows the overall workflow of the method. In (a) the bayesynergy R package [20] is used as a pre-processing step to estimate the latent GP underlying each experiment, which is summarized by its posterior mean and variance. In (b), the posterior mean and variance of each experiment is put into a common grid, and collected in matrices \({\textbf{Z}}\) and \({\textbf{S}}\). In (c), the missing entries of \({\textbf{Z}}\) is predicted using the PIICM model. And finally, in (d) the predicted latent GP of an unseen experiment is used to reconstruct the dose–response function