Question of which F combinations are most likel given evidence from temporal data? Use of fitting regulation models to address this - Shannon McWeeney

Fitting a single TF linear activation model (concentraton based) - goal is to fit genome wide - Shannon McWeeney

Challenges - model parameters unknown ; few time points; noisy data; "open" system because of f(t) - concentration of tf. Their solution: can model f(t) as Gaussian process - Shannon McWeeney

examples of approach from Honkela et al PNAS 2010 - Shannon McWeeney

interesting discussion on how to evaluate using chIP-seq data - Shannon McWeeney

Discussion of new extensions - multiple TF - Shannon McWeeney

examination of enrichment of pairs (TF-target link and TF-TF) - Shannon McWeeney

noted that TF activated by signalling can be modelled as latent - Shannon McWeeney

tigre bioconductor package - Shannon McWeeney

could incorporate chIP-seq data for priors (with-held here for evaluation purposes) - Shannon McWeeney

Q: could you adapt model to focus on TF that spike - influence behavior of other TF quickly? A: could adapt via binary switches - Shannon McWeeney