search

RGLab / openCyto

A package that provides data analysis pipeline for flow cytometry.
GNU Affero General Public License v3.0
77 stars 29 forks source link

gating on multiple populations with flowClust.2d #210

Closed browshanravan closed 4 years ago

browshanravan commented 4 years ago

My understanding is that openCyto:::.flowClust.2d function recognises K user defined clusters (as visualised by the argument plot=TRUE) however it only gates on the one with the most cell count if the target argument is not defined.

Is there a way to gate on more than 1 population, say 2, and have them added to the gate path as myGateName_Gate1 and myGateName_Gate2?

mikejiang commented 4 years ago

The builtin flowClust.2d in openCyto package is designed for returning only one gate/population as you've already noticed.

To return multiple gates from one gating function, you will need to wrap the flowClust into your plugin function. Here is one example https://rpubs.com/wjiang2/194332 that demonstrates this Be aware that it is just proof-of-concept code written years ago, (e.g. tmix2DGate has been moved to https://github.com/RGLab/cytoUtils/blob/master/R/tmix2DGate.R) try get the principle and tweak into your own needs.

browshanravan commented 4 years ago

Everything works! great! thank you! I was hoping to ask two short questions which are unrelated. Should I open a new issue?

browshanravan commented 4 years ago

Having used your custom scripts linked above, I have successfully identified my 2 populations of interest, as shown below.

Screen Shot 2020-05-01 at 03 33 25

I have achieved this by running the below script.

thisData<- gs_pop_get_data(gs, "parent_pop") DD_gate<- fsApply(thisData, function(fr) tmix2DGate(fr, channels = c("chnl1", "chnl2"), K=2, quantile=0.95)) gs_pop_add(gs, DD_gate, parent = "parent_pop", name = c("gate1","gate2")) autoplot(gs, c("gate1","gate2")) recompute(gs)

However, I have recently realised when I select one of the gates for plotting using autoplot(gs, "gate1"), the gates seem to be assigned randomly throughout my data, as shown below.

Screen Shot 2020-05-01 at 03 32 08

Given the arrangement of my clusters, I though of using the target = c(0,0) argument in order to constrain one of the gates, say gate1, on the x and y axis to consistently recognise the population/cluster closest to the origin but I was not been successful. Do you have any suggestions as to what I can do?

gfinak commented 4 years ago

Hi, @browshanravan sorry your issue seems to have been missed. I'd like to dig into this, can you share a little bit of the data with code to reproduce this? Thanks

mikejiang commented 4 years ago

I don't think target parameter works for tmix2DGate, you will need to switch back to flowClust.2d to gate one population at a time.

mikejiang commented 4 years ago

For now, I will close it with https://rpubs.com/wjiang2/176974, which illustrates the usage of target with flowClust.2d. Feel free to reopen it if you feel the issue is not fully resolved.