Analyze_integrated with a integrated seurat object

[vincentvfdg]

• For analyze_integrated, if the seurat object in input is an integrated one, we should put assay = "integrated". If assay = "integrated", the filtration step (aka step 6), should be skipped. Otherwise, it will look at nCount_integrated, which does not exist.
• I modified the FindNeighbors function in Seurat.R by adding graph.name = "RNA_snn". Usually, this is the base parameter for graph.name, but I think, when using integrated seurat object, the graph.name automatically changes, which disrupt the flow for FindClusters afterward. By still specifying RNA_snn, the flow is working. We could modify this parameter in the future by making an ifelse loop that allows an integrated seurat object to have a different name (and set that name for us), which should also be applied to FindClusters.
• Maybe it is a small detail that needs further small discussion, but having one single report repository for two separate analyses feel a bit heavy. Same things apply for the plot repository.
• If the dataset is very big (for example ESTJ20 with a size of 3.2 GB after integration), we should totally save a seurat_object per step, not just the 12th one. This particular analysis takes about 20 minutes just to reach step 12, which is considerably longer than most of the projects so far. We probably won't be working with larger dataset, but we never know, and it would be VERY helpful to have this setting implemented.