Fig. 7From: PureCLIP: capturing target-specific protein–RNA interaction footprints from single-nucleotide CLIP-seq dataSummary of the hidden Markov model framework. a Starting from mapped reads (bottom), observations are deduced (individual read start counts, which are further smoothed to obtain pulled-down fragment densities) and combined with additional covariates (top) to reconstruct the most likely sequence of hidden states (middle). b Graphical representation of the corresponding non-homogeneous hidden Markov model. CL crosslink-associatedBack to article page