Question

singlecellexperiments markers after clustering

0

Entering edit mode

achaillon • 0

@achaillon-14117

Last seen 4.0 years ago

Hello I am using the CATALYST::cluster function on my flow data. My panel includes 13 markers

panel$fcs_colname [1] "Alexa Fluor 700-A" "PE-CF594-A" "APC-H7-A" "BV421-A" "BV711-A" "BV786-A" "BV605-A" "PE-Cy7-A" [9] "BV650-A" "APC-A" "PE-A" "PerCP-Cy5-5-A" "Alexa Fluor 488-A"

and are present in my flowset object

all(panel$fcs_colname %in% colnames(fs1)) [1] TRUE

but after running

CATALYST::cluster(sce, features = NULL, xdim = 10, ydim = 10, maxK = 20, seed = 1234)

I only have 11 markers left....

DataFrame with 11 rows and 4 columns channel_name marker_name marker_class used_for_clustering CD3 Alexa Fluor 700-A CD3 type TRUE CD4 PE-CF594-A CD4 type TRUE CD8 APC-H7-A CD8 type TRUE HLA-DR BV421-A HLA-DR type TRUE CD38 BV711-A CD38 state TRUE CCR7 BV786-A CCR7 type TRUE CD45RA BV605-A CD45RA type TRUE PD-1 PE-Cy7-A PD-1 state TRUE TIM-3 BV650-A TIM-3 state TRUE CD73 PerCP-Cy5-5-A CD73 state TRUE CD160 Alexa Fluor 488-A CD160 state TRUE

I am probably missing something obvious... can you help?

thank you!

CATALYST SingleCellExperiment • 1.3k views

ADD COMMENT • link updated 4.0 years ago by helucro ▴ 70 • written 4.0 years ago by achaillon • 0

0

Entering edit mode

Hard to read the output above - but if I'm not mistaken, the missing channels are APC-A and PE-A. These are the only ones not containing any numbers. So my guess is this:

By default, prepData() will retain only mass channels in the assay data. All none mass channels (e.g. in CyTOF, these are time and event length) are moved to the int_colData().
Thus, when you construct the SingleCellExperiment, the object already has only 11 features, and this is not the result of running cluster().
Now, since this is FACS data, there are no mass and no-mass channels, and we want to retain all parameters.
To enforce this, you should run prepData(..., FACS = TRUE). All 13 markers will then remain in the main assay data.

ADD REPLY • link 4.0 years ago helucro ▴ 70

0

Entering edit mode

I confirm it works with FACS=TRUE. thank you so much. I will follow with another question but using a different thread

thank you

ADD REPLY • link 4.0 years ago achaillon • 0

score 1 · Answer 1 · 2020-12-20

Solution:

Run prepData() with argument FACS = TRUE.

Explanation:

By default, prepData() will retain only mass channels in the assay data. All none mass channels (e.g. in CyTOF, these are time and event length) are moved to the int_colData(). Since this is FACS data, there are no mass and no-mass channels, and we want to retain all parameters. Setting FACS = TRUE when running prepData() to construct the SingleCellExperiment enforces this.