Thank you so much! This has been really helpful :) It was easy to follow. And I like that you explained the individual parts (tsne, FlowSOM) before showing the whole process merged together.
Great tutorial! I have 25 samples (mouse lung, extracellular) - 5 mice per group, 5 groups (Mock Mock , Mock Infected, Infected Mock, and so on). Should I concatenate on only 1 sample from each group and have that as my merged file as a "representation" of everything and THEN run tSNE (thus concatenate 5 sample) or should I merge all 25 samples? The other question is should you ALWAYS concatenate in these instances before running any tSNE ?
I think the safest way is to concatenate on everything, otherwise you'd lose any statistical power. You'd effectively only have n=1 per group. However, that being said, this workflow starts to become pretty impractical when you have so many samples! It's one of the main downfalls I've seen. For example, let's say you wanted 100,000 total events to run tSNE. If you concatenated all, then you'd only have 4,000 cells per sample - a very small number to figure out population frequencies from each sample, especially if you're trying to gate rare populations. If you do separate tSNE runs on separate samples, the same cell populations wont necessarily plot to the same places on the map, thus, you couldnt compare the tSNE maps to each other. I'd try concatenating a much larger number of cells so you make sure you're getting an acceptable number per sample and running tSNE on everything. But it may take a really long time to run...
Great seminar! Thank you! A question I have is the following: you concatenated and run tSNE and FlowSOM and then dragged FlowSOM to the gated 4 samples. Is not possible to upload your four samples (with the total number of cells recorded) in the same workspace and drag the FlowSOM analysis from the concatenated sample on those four samples samples that now have all recorded data?
Thank you for this question. Ideally you would be able to do exactly this and it would make your life easier and your data more robust. However, I have never been able to do it in practice. If you try to drag the flowSOM gates onto the original files it will tell you there is a missing parameter and not allow it. If you try and run flowSOM on one of the original files and select "Apply on map" for the flowSOM map generated on the concatenated file, the execution halts in R for an unknown reason to me. So if anyone can figure this out it would help me as well!
Thanks for the tutorial! Very clearly explained! One question about comparing FlowSOM population frequencies. Let's say I have a control and a treatment group, with 5 samples in each group. I create a single concatenated file with all 10 samples, then separate each sample using sample ID gates, and then run FlowSOM on that whole concatenated file. Then I drag and drop the created FlowSOM population gates onto each one of the Sample ID gates. My question is: Can I then run a simple two-tailed t-test to compare the frequency of a given flowSOM population between control and treatment samples? Would that be statistically valid/"legal"?
Hi! Thank you for the great tutorial! For some reason if I open anything in layouts the axes lose the number there... do you by any chance know how to change that?
![FlowJo [tSNE]](http://i.ytimg.com/vi/UK725D14-bo/mqdefault.jpg)
Thank you so much for this tutorial! Really well explained and easy to follow!
Thank you so much! This has been really helpful :) It was easy to follow. And I like that you explained the individual parts (tsne, FlowSOM) before showing the whole process merged together.
It is a great tutorial! Well explained, moving with good speed and presented in a very logical fashion. Thank you so much!
Thank you for this really good tutorial :)
Great explanation of the workflow, really easy to understand. Thanks!
Great tutorial! I have 25 samples (mouse lung, extracellular) - 5 mice per group, 5 groups (Mock Mock , Mock Infected, Infected Mock, and so on). Should I concatenate on only 1 sample from each group and have that as my merged file as a "representation" of everything and THEN run tSNE (thus concatenate 5 sample) or should I merge all 25 samples? The other question is should you ALWAYS concatenate in these instances before running any tSNE ?
I think the safest way is to concatenate on everything, otherwise you'd lose any statistical power. You'd effectively only have n=1 per group. However, that being said, this workflow starts to become pretty impractical when you have so many samples! It's one of the main downfalls I've seen. For example, let's say you wanted 100,000 total events to run tSNE. If you concatenated all, then you'd only have 4,000 cells per sample - a very small number to figure out population frequencies from each sample, especially if you're trying to gate rare populations. If you do separate tSNE runs on separate samples, the same cell populations wont necessarily plot to the same places on the map, thus, you couldnt compare the tSNE maps to each other. I'd try concatenating a much larger number of cells so you make sure you're getting an acceptable number per sample and running tSNE on everything. But it may take a really long time to run...
Hi, did you succeed in doing this cross samples comparison? thanks
Yes!! I did complete it!! And I did it as suggested 😊
@@Mariiamdov Thanks. So did you merge the 25 mice into one sample and separated them by the "sample ID" keyword?
Great question! I have a similar problem but with 51 samples. how did you separate your samples on the plot after concatenation?
Many thanks for this tutorial, very helpful.
This was very well explained, thank you ❣️
I can not see tSNE plot option when I open the FLOWSOM window, I don't know what I am missing. Can anyone help?
Great seminar! Thank you!
A question I have is the following: you concatenated and run tSNE and FlowSOM and then dragged FlowSOM to the gated 4 samples. Is not possible to upload your four samples (with the total number of cells recorded) in the same workspace and drag the FlowSOM analysis from the concatenated sample on those four samples samples that now have all recorded data?
Thank you for this question. Ideally you would be able to do exactly this and it would make your life easier and your data more robust. However, I have never been able to do it in practice. If you try to drag the flowSOM gates onto the original files it will tell you there is a missing parameter and not allow it. If you try and run flowSOM on one of the original files and select "Apply on map" for the flowSOM map generated on the concatenated file, the execution halts in R for an unknown reason to me. So if anyone can figure this out it would help me as well!
have you found a solution to this yet? @@ucmercedscifstemcellinstru711 loved your tutorial and found it very useful!
Reaaaaally helpful video. Thank you so much.
Thank you for this great tutorial! It helped me a lot!
Thanks for the tutorial! Very clearly explained!
One question about comparing FlowSOM population frequencies.
Let's say I have a control and a treatment group, with 5 samples in each group. I create a single concatenated file with all 10 samples, then separate each sample using sample ID gates, and then run FlowSOM on that whole concatenated file. Then I drag and drop the created FlowSOM population gates onto each one of the Sample ID gates.
My question is: Can I then run a simple two-tailed t-test to compare the frequency of a given flowSOM population between control and treatment samples? Would that be statistically valid/"legal"?
Great question.. wouldlove to know this too
I have the same question!
Thank you so much. That's very helpful!
Thanks! Great tutorial
Great tutorial, thanks a lot!
What if tSNE and FlowSOM do not cluster the same?
Hi! Thank you for the great tutorial! For some reason if I open anything in layouts the axes lose the number there... do you by any chance know how to change that?
right click on thr graph and go to properties, then unclick hide axes/labels/numbers etc. Hopefully you found this already