HectorRDB
commented
4 years ago Hi, Thanks for running tradeSeq, I'm glad you like it.
For now, the monocle3 / tradeSeq part is very experimental since monocle3 itself is still in its beta version. So the code in the monocle / tradeSeq vignette is resorting to two hacks:
- in the first chunk (where you have the error), we try to identify the root node that is closest to our starting cluster.
- in the second chunk, we extract the lineage assignements for each cell.
So now, to your issue. I would guess that the problem is that there is no root point (i.e a terminal node in the graph) in that cluster. That should appear when running those lines
closest_vertex <- paste0("Y_", closest_vertex)
root <- names(which(igraph::degree(principal_graph(cds)[["UMAP"]]) == 1))
root <- root[root %in% closest_vertex]There is not intersect between the roots and the closest_vertex.
This code is inspired from https://cole-trapnell-lab.github.io/monocle3/docs/trajectories/, the get_earliest_principal_node function.
I would recommend either
- Tweak the graph construction until there is an overlap or
- opening an issue with monocle3.
Let me know if this is clear.
Seandelao
Hello,
Thank you for the wonderful package, I am very excited to utilize it for my data sets!
I have created a cds object from an integrated Seurat object, but an having issues with calculating pseudotime by following the code provided by the tradeSeq tutorial. I can calculate it fine following Monocle3's method in their tutorial, but then I don't know how to modify the tradeSeq tutorial to extract the pseudotime and cellweight values.
When I run: exprs_human<- GetAssayData(Alpha_Beta, slot="counts", assay = "RNA") counts <- as.matrix(exprs_human)
meta_data <- Alpha_Beta@meta.data keep <- ("CellfindR") meta_data <- meta_data[keep] names(meta_data)[names(meta_data)=="CellfindR"] <- "cellType"
df <- data.frame(cells = colnames(counts), cellType = meta_data) rownames(df) <- df$cells
cds <- new_cell_data_set(counts, cell_metadata = df, gene_metadata = data.frame(gene_short_name = rownames(counts), row.names = rownames(counts)))
cds <- preprocess_cds(cds, method = "PCA") cds <- reduce_dimension(cds, preprocess_method = "PCA", reduction_method = "UMAP")
plot_cells(cds, label_groups_by_cluster = FALSE, cell_size = 1, color_cells_by = "cellType")
umaps = Embeddings(Alpha_Beta, assay="integrated",reduction="umap") reducedDims(cds)$UMAP <- umaps plot_cells(cds,color_cells_by="cellType")
cds <- cluster_cells(cds, reduction_method = "UMAP")
cds <- learn_graph(cds) plot_cells(cds, label_groups_by_cluster = FALSE, cell_size = 1, color_cells_by = "cellType")
celltype <- as.character(meta_data$cellType) cell_ids <- which(celltype == "3.1.0") closest_vertex <- cds@principal_graph_aux[["UMAP"]]$pr_graph_cell_proj_closest_vertex closest_vertex <- closest_vertex[colnames(cds), 1] closest_vertex <- closest_vertex[cell_ids] closestvertex <- paste0("Y", closest_vertex) root <- names(which(igraph::degree(principal_graph(cds)[["UMAP"]]) == 1)) root <- root[root %in% closest_vertex] cds <- order_cells(cds, root_pr_nodes = root)
I get the error: Error in data_matrix[, ((((i - 1) * block_size) + 1):(ncol(data_matrix)))] : subscript out of bounds
Everything seems to be working well until I run the line: root <- root[root %in% closest_vertex]
in which it seems to return an empty character string.
Any idea what might be happening? Or, alternatively, how I might be able to extract the pseudotime and cell weights from the cds object after selecting the root cell by the current method suggested by Monocle3's tutorial? I don't have the data 'root' if I use their method...
Sorry if that was confusing, happy to clarify! Thanks again.
-Sean