Hello. In this video, I'm going to discuss how to transform your data displays using the T transform button, which is located in the graph window next to the access labels, where you can change the parameters that are displayed on the graph plot. There is a T button right next to that, and it allows you to change the graph scaling and change the way the data is displayed here in the graph plot.
Most fluorescent cytometry data is going to be best displayed on what we call a "biexponential transform function." A biex transform means you have a linear range around zero, so that you can see the negatives, the negative events that are not expressing your marker of interest. Then a log range further out, as you get away from zero. It turns into a log scaling, so you can see the 10- or 100- or 1,000-fold difference in biological expression of your marker.
With the T button here, we can go to the customize axis dialogue, and it will open up this transformation window, and the transformation window shows a histogram profile of the actual marker that you initiated the T button from. In this case, the CD3 distribution is as shown here. It's nice and bi-modal, and I'm going to show you a few different options here. Right now, this scaling looks pretty good to me. I can see all of the events on scale. Nothing's crushed up along the axis. It's not compressed too much around the zero point in the linear range, but what happens if you bring in data and it looks more like this? Totally crushed on the axis, where you're missing half of the events associated with your negatives ...
Well, there's not enough backend negative space being displayed here to see all those negative events, and so you go to the T button, initiate the customize axis option here ... There are a couple different options. You can add or subtract space on the back end by using this extra negative decade slider bar to move all of the data over so you can see the negatives and get them off the axis. You can add or subtract space on the top end by using the plus or minus buttons next to the parameter name here. That will just add a space on the top end, show you a different scale if you have an instrument that collects at 10 to the seven.
A couple of plus-button clicks will bring that data on-scale. Then you can change how much of the visual space of this axis is given to the linear range, scaling around zero or the log range scaling further out using this width-basis slider. If you go to right to a larger negative number, you'll see the number going up. It's a negative 316 here and that it compresses the data around zero because going to the right, a larger negative number for the width basis gives more visual space to the log scaling area and less to the linear. It compresses all that data around zero.
If you go to the other direction, go to the left with the slider bar, it spreads the data out around zero. Using these different options, I can make the data look extremely compressed like I've done here, and I can apply this change by clicking the apply button. The change in scaling can be applied to more than one parameter. Even though I set the scaling here, I can click on the other parameter that I'm displaying in this graph window, and when I apply, both parameters will be scaled and get the same transform values that I've set up there.
In this case, this is not an appropriate way to display your data. If you want to see all the data and know where to draw your gates, you need to actually display it within the graph plot. This is way too compressed on the axis for me and my liking, but if I go the other direction, I can give a lot of extra negative decade space and compress all the data around zero. Then when I press apply and apply those transforms to both of those parameters, obviously, I'm wasting a whole lot of space here. The data is all compressed at scales, and I can't see where I would draw my gates. It would really not be a very good plot to start gating and using those tools on.
What I'm looking for is a nice, happy medium here, somewhere in between the two extremes. Usually, I find that a negative 10 width basis is a good place to start when you're looking for nice distributions. Then all I have to do is move the data off the axis a little bit with a little move of extra negative decade space. Applying that change gives me a nice distributed plot in which the visual medians of these distributions are consistent with the actual statistical medians. I can see all the data. It makes this high resolution, flow cytometry data more appealing to the eye than to have it compressed around zero or piled up on the axis. You're looking for that nice sweet spot that you can show all that data in a usable fashion.
Clearly, there are other options besides biexponential transformation, but most fluorescent cytometry data will be best displayed on this biex scale. If I go to just a linear scale, you can see that's not really appropriate for this distribution. If I go to a log scale, I miss half of the negatives that are below zero because a log scale will only bring me down. You don't see anything that's negative, and if you go to customize axis, there are alternative scales here.
I can go back to my biex scaling, set it as I want and apply the change to both parameters I have, so I can see all my data. There is this scale drop-down list that I can change to linear or log, and there are also additional transformation types that may be useful, depending on the instrument that you've acquired off of, arcsine if you have a CyTOF instrument. It may be a Miltenyi transform if you have a Miltenyi product.
By and large, the best distribution and way to display that data is going to be on this biexponential transform function, where you can change the top-end scaling, change the bottom-end space and change the compression around zero so that you get a nice, pretty plot to share with your group or put in a publication. This is Tim with FlowJo. Thanks for sticking with me on this little talk about transforms, and please stay with us for some additional features as we go through this series of videos. Take care