Hello I am using SingleCellExperiment workflow for my flowdata and it works really well. I am trying to use an alternative clustering method (Phenograph) to identify the clusters, but as you know phenograph output are csv: it consists on one large summary csv without cluster info and separate csv (one per cluster) I did try to convert back to flowset using
counts = as.matrix(pheno_df); sce =SingleCellExperiment(list(counts=counts))
it gave me
sce class: SingleCellExperiment
dim: 25000 13 metadata(0): assays(1): counts
rownames: NULL rowData names(0) colnames(13): APC-A APC-H7-A ... PE-Cy7-A PerCP-Cy5-5-A
colData names(0)
reducedDimNames(0)
altExpNames(0)
summary(sce)
[1] "SingleCellExperiment object of length 25000 with 0 metadata columns"
I can retrieve sample_id and cluster_id from the csv sources but the dimension is not ok
should rather be something like:
"SingleCellExperiment object of length 13 with 4 metadata columns"
with experiment info being sample_id condition patient_id n_cells ,etc...
Is there a tutorial to go import phenograph output and metadata to the SingleCellExperiment workflow
thank you
Rphenograph, which takes as input an expression matrix and returns (among other things) a vector of cluster assignments. Otherwise, I am assuming the .csv also contains or lets you retrieve some sort of cluster assignment, in which case:CATALYSTframework (see here). By following the steps exemplified in there, it is in principle possible to use any clustering method. Briefly, this requires i) modifying thecluster_ids stored in thecolData, and ii) modifying thecluster_codesin themetadata.thank you! I will give it a try