--Brian has implemented scripts to compute GRNs, score each cell for the activity of GRNs, created tSNE visualizations of these networks with Scanpy, which look very good, and calculated enrichment of networks in cell types.
--I think we could define tasks for the next sprint that would help us think about whether SCENIC could/should be implemented within gEAR and, if so, what that would look like. For instance, it's definitely feasible to implement cell scoring based on pre-computed GRNs using the R package AUCell or a port of that package into Python.
** At this time, SCENIC is very computationally intensive - for datasets with ~5000 cells, the process takes about a day using 16 cores and 100G of ram. I'm in favor of the pre-computed route. Also, there are some functions from AUcell already written for python, but I don't know if all functions have been ported over yet.
--Brian has implemented scripts to compute GRNs, score each cell for the activity of GRNs, created tSNE visualizations of these networks with Scanpy, which look very good, and calculated enrichment of networks in cell types.
--I think we could define tasks for the next sprint that would help us think about whether SCENIC could/should be implemented within gEAR and, if so, what that would look like. For instance, it's definitely feasible to implement cell scoring based on pre-computed GRNs using the R package AUCell or a port of that package into Python.
** At this time, SCENIC is very computationally intensive - for datasets with ~5000 cells, the process takes about a day using 16 cores and 100G of ram. I'm in favor of the pre-computed route. Also, there are some functions from AUcell already written for python, but I don't know if all functions have been ported over yet.