Figure 5

Bayesian hierarchical model for protein colocalization. This diagram illustrates the Bayesian hierarchical model employed in ProteinCoLoc to assess the correlation between pixel intensities within patches of image I. The likelihood is calculated using a Student T-distribution parametrised by the mean correlation coefficient \(\overline{{\rho_{I} }}\), the dispersion parameter \(\sigma_{I}\) and the degrees of freedom \(\nu_{I}\). These parameters for individual images are sampled from global hyperparameters \(\rho_{S}\), \(\sigma_{S}\), and \(\nu_{S}\) for images within the sample condition or from the global hyperparameters \(\rho_{C}\), \(\sigma_{C}\), and \(\nu_{C}\) for images from the control condition.