We decided to create a video that would show the VenturiOne Discovery and CytoSwarm platforms working together, performing multidimensional analysis on a simple lymphocyte subset file to walk everyone through the basics of getting to know your data and how to associate what conventional flow analysis shows us with what the algorithm data shows.
Have a look at the video and let me know your thoughts.
Please remember to like the video and subscribe to the Applied Cytometry YouTube channel to stay up to date on our new content.
There is also a written version with illustrations below.
I wanted to create a walk through demonstration of how easy it is to create and look at high dimensional data using the VenturiOne Discovery platform.
To keep things simple, I am using a single lymphocyte subset FCS file for high dimensional analysis.
So, opening the CytoSwarm application,
I select the file I want to run, change the job name to what I want.
Press next, then select the algorithms I want my data running through, so here I will use UMAP, t-SNE and FlowSOM.
Next, I de-select the parameters I don’t need to use for the analysis.
Pressing next here would lead me through the individual settings for the selected algorithms, but I am going to use the default settings here so I press finish.
….and now you see the file is processing, so we just wait for that to complete.
When the analysis is complete, if you highlight the job and press the details button you can see the rough shapes of the algorithm data as you click through them, if you have had any errors reported you would be able to find out here which algorithm caused the error and within the text box find out what the error was.
I close this and go back to the highlighted job, now if I want to download all the steps individually I can do using the download all steps button, but I want to see all the algorithms together, so I will download the results only and save it to my specified location.
I have VenturiOne Discovery open and ready to use, so I will now load in my newly saved file. As this is a simple T cell subset file, I will quickly gate the data, then we can have a look at the multidimensional data.
I change the default FSC/SSC plot to a SSC/CD45 plot and put an autogate around the lymphocyte population, I then create two more dot plots showing CD4/CD8, and CD16 and CD56/CD19 and gate these plots on Gate A (Lymph gate) then create gates for the following, as can be seen in the picture above: CD4 pos, CD8 pos, CD4 and CD8 pos, CD16 and CD56 pos, CD19 pos.
With the basic gating complete we can now go to the Discovery tab of the ribbon, we need to set the ‘gate by’ to A and the ‘tree by’ to flowsom.
The cluster list tab shows you all the clusters the algorithm created and the event count in each, then the nearby columns shows the nearest cluster ID number and the relative distance to the parent cluster.
Moving to the Cluster Tree tab it displays the FlowSOM plot of the data, change the tree layout (as many times as you need, you can also undo) to best represent the gated data.
The reason I did a little conventional gating is so I can use those gates to show where the data is on the flowsom tree by back gating the data and allowing me to create and colour the cluster tree.
So Gate B represents my CD4 population which is coloured the same as the gate colour.
I now know where the green population sits so I can stop the back gating, create a cluster group in the same colour and fill in the correct clusters in the tree, and I do this for the four main subset gates (B,C,E,F) and show them all together on the Cluster Plot that I have created.
You can see from the Cluster plot that we only have a few clusters left that have not been densely coloured using these gates, so I will add in the UMAP plot to have a look where these populations lie and see if they have been separated clearly.
I have gated the UMAP plot on gate A, and here we have the lymphocyte populations.
I also added a gate (gate G) to the umap plot and use that gate to back gate on to the flowsom plot or on to other plots.
The picture above shows where these populations lie on the flowsom plot, this is powerful imagery as it lets you relate how the data is separated using these algorithms.
There are some red events just below the NK cells on the UMAP plot and if I include these in gate G they show up where I hadn’t been able to confidently colour on the flowsom.
With these events in mind I will now back gate on the dot plots and see where these cells are sitting.
and it is showing these events in the negative area of both plots suggesting that these may not be lymphocytes or typical lymphocytes.
looking further into this, these cells definitely fall within the lymphocyte gate A, you can see them batter against the ungated data, and I will change the gate colour to a brighter colour to make it stand out.
Feel free to comment and tell us what you think these might be and why!
We can also do the same on the tSNE data, so I back gate the flowsom on gate H and as I move the gate around it shows where those populations lie on the flowsom plot.
So this has purposefully been a nice simple file to look at, where the populations are nicely separated by the VenturiOne software and the different algorithms,
I believe that this will help develop an understanding of how these results are relatable and how we can bridge the gap between old style single and dual parameter analysis and the algorithm clustered data that is becoming evermore popular as the demand for more data and more parameters increases, with these being the new way to visualise our data it is vital that we also have a way of relating the clustered data back to what we know well.
Thank you for reading.
Please like the post and share if you found this informative and interesting, and please visit the Applied Cytometry website for more information and free trials of the VenturiOne and CytoSwarm software.
keep a look out for the next post in which i will be looking at how to analyse multiple files.