Mpath expects landmarks to be ordered with prior data

[rcannood]

In the example scripts, the nbor ordering is done by manually inputting the order of the landmarks:

#' ### path 1: "T0_2","T0_1","T24_8","T48_10","T72_13"
#' 
#' ccFile = "GSE52529_fpkm_matrix_nooutliers_ANOVA_p0.05_DEG_landmark_cluster.txt";
#' order <- nbor_order(exprs = "GSE52529_fpkm_matrix_nooutliers_ANOVA_p0.05_DEG.txt",
#'                     ccFile = ccFile, 
#'                     lm_order = c("T0_2","T0_1","T24_8","T48_10","T72_13"),
#'                     if_bb_only=TRUE,
#'                     method=1)

We should think about whether we should also allow to provide the ordering of the cell grouping as prior.

[zouter]

Are you sure this ordering is important? Hmmm this can explain why the performance is low so frequently. The ordering of the cell grouping is given in the state network, so we can just use that one, although in principle this is not necessarily the case, so I am going to add a "group_network" to the tasks, ok?

[zouter]

Is this still an issue now? I don't see where I can input the ordering to nbor..

[rcannood]

Example from Mpath.R:

order1 <- nbor_order(exprs = "GSE52529_fpkm_matrix_nooutliers_ANOVA_p0.05_DEG.txt",
                     ccFile = ccFile, 
                     lm_order = c("T0_2","T0_1","T24_8","T48_10","T72_13"),
                     if_bb_only=TRUE,
                     method=1)

Here, the lm_order is used to order the clusters. Is this what you were looking for?

[zouter]

Yes, but the nbor_order function is never called in the wrapper!