Hi Safa! Thank you so much for reaching out. Unfortunately I personally don't have experience with the BD FACSLyric flow cytometer. When you ran the single color controls can I ask if you made sure you used the same antibody fluor conjugate as what was in your panel? Did you check to make sure that the single color was as bright as your experimental sample? The CS&T passing performance check is great, but I'm more curious about how the single colors were compared to your full stained sample.

Thanks for your question and feedback. It really makes it worthwhile to develop these initiatives when we see it makes a difference to those watching these videos. As for your question, yes, this is a well known problem with Diva that was never corrected. BD expects their customers to find workarounds for bugs and badly implemented features in Diva. Unfortunately, most of us learnt the hard way, after losing important samples due to this issue.

@@RuiGardner thanks for your answer. Your work is much appreciated! So, do you suggest that we first set all gates with biexponential scale to better see our populations, and then change to log scale when we start to sort? But in this case the gates will still go below zero also on the log scale (if we are sorting a negative population of course). I have never noticed that I have lost cells due to the use of biexponential scale. How can I prove this on my instrument? I think biexponential scale it is almost always used in our instrument.

@@CitometriaCPTUnvr This only happens if you're using biexponential AND all your gate is below zero for one of the parameters. Switching to log scale may be difficult because you may have all your data squashed in the lowest bin. What we typically do is to try and have at least some of the gate above zero, and if that's not possible (maybe the gate is already "touching" the positive populations) we increase the gains to have part of the negative population above zero. This way the gate will also be above zero. You can test this by sorting in biexponential with very low gains so that your negative population is below zero. Then make sure the gate is also below zero and see what happens. It may vary with different Diva versions (although the latest still show the same issue)

@@RuiGardner thanks I will try it. But in your experience, what happens? Does the DIVA software not recognize the cells below zero and therefore the events are aborted? Thanks again for the explanations!

Hi! Thanks for the question. In this case, Kathy wasn't trying to verify what was cut off in the singlet gating, but the opposite. She was verifying what wasn't excluded. To be more precise, that dimmer positive population she gated happened to be some aggregates that hadn't been excluded. But visualizing the colored population in the doublet exclusion plot she saw that we were not gating out all the doublets/aggregates. Once she removed those from the singlets gate, that dimmer population (P4) disappeared. A nice way of using backgating to verify if we have cleaned all our data appropriately. It's one of several tools to improve the quality of the data.

Thanks for the kind words, so glad the video was helpful. I'm not sure exactly where you got confused, but one of the points was that if you assume the drop-charge delay (timing between laser and breakoff point) if you have two cells attached and you don't exclude them, the following scenarios can happen: 1) A negative and a positive cell stuck together will have the total signal of a positive cell (with a little bit more signal from the autofluorescence of the neg cell). So when you sort for positive cells and you don't exclude doublets, those two cells will be sorted together and may split when collected in the tube and you will detect that negative cell when you read the sorted sample back in the instrument. 2) if two negative cells or two positive cells are attached together, then you'll sort those in the respective gates and they won't count for impurities. So if you see a positive cell in your collection tube when you were sorting the negative population, then it can't be because of lack of doublet exclusion because the doublet with a positive cell would fall in the positive gate in the first place. Did this answer your question?

You want to find the minimum amount of antibody that will give you maximum separation. The less antibody you use, the more money you save and you also reduce the likelihood of any unspecific staining. But you need enough to maximize the separation between negative and positive populations for that specific marker you're measuring. Sometimes, however, if you already have a really good separation, and you are not looking at differences in levels of expression, you can reduce the amount of antibody to minimize any potential spillover signal of that conjugated fluor in other channels. You still end end up a good separation, but you reduced the amount of signal that can potentially spillover into other channels.

We used the Slingshot particles to demonstrate the principles of compensation. These principles will be the same for any other tissues but the particles themselves are suitable for blood markers only.

The FITC channel as well as the PE channel of the 488 laser are typically good channels for AF. There are also channels in the 450-550nm emission range off the violet or UV channels that will also be good to check AF. This obviously depends on the AF spectral signature of your cells, so trying out a few channels where you see an obvious diagonal shape in the unstained population(s) will do the trick. But definitely not random :)

Hello! Kathy here. I do understand what you are saying here in terms of lowering individual detector gains. I think from an end user perspective there is a danger with starting to adjust individual gains. If you are an SME and want to play around with individual gains vs. gain reduction for the whole laser line, go for it and see what yields best results! For a user of the technology that is learning, I think there is a lot that can go wrong with individual gain reduction, especially as you are not accustomed to this workflow. In the case of DAPI, it is usually screaming screaming bright on UV. If you are in a scenario where you can't titrate down the concentration of the dye being used, because it is so bright off of so many detectors, I am in favor of reducing the full set of ADPs. I do understand what you are saying though! It would be fun to test out a whole of different scenarios. :-)

This is a great question, to which I don't have a definitive answer to, as we haven't tested different scenarios in a systematic way. But think of it this way. With spectral flow cytometry the fluors are identified and even distinguished based on the proportions of signal between each detector, i.e., their spectral signatures. Let's say you have two red dyes, F1 and F2. The first one peaks in the first red channel (R1) then drops almost to zero in R2 and the other is the opposite. If you reduce R1 gain significantly, you din't really impact the signature of the second fluor (F2), but you did impact significantly the capacity to measure F1. And maybe someone can correct me if I'm wrong, but in this scenario I would say not only you limited the chances to quantify F1 (you reduced significantly the channel that was providing most of the information to detect F1), but you've also changed the similarity between them (to better or worse). So it also makes it hard to predict whether a panel will work or not if you later change those proportions. I may be wrong, but without knowing enough, I would suggest maintaining the proportions/signatures and just changing all the channels proportionally rather than playing around with the gains individually.

Hi Matthew, that's a very common question. I guess the way to think about it is, if there's anything that can affect your data, you need to control for it. You are correct, fixation can reduce the brightness of fluors. This won't be a problem per se, unless the loss of brightness is such that you no longer have a good separation. But in this case, you would also see that kind of loss in your sample. Since we need to make sure that the signal in the beads is brighter than in the sample, I would be more worried about loss of resolution in the sample. The biggest problem is that fixatives can degrade (or at least change) the spectral properties of fluors, for instance, through changes in pH, crosslinking of peptides, etc. So it would be ideal to have the beads go through the same fixation protocol, including how long they will be exposed. Having said that, if from your experience, you don't see any changes in the amount of compensation/unmixing in your samples when using "fixed" beads made on the same day vs made on the day you fixed the samples, then you can start making them on that same day. It may depend on the fluor, so I would test all of them a few times to give you confidence. But remember, every time you drop a control, you're no longer in control! ;)

Hi Ramesh! Of course, it is our pleasure. I typically suggest users to turn on all lasers, but you do have the option to turn on the lasers from the experiment setup window if you have forgotten to turn it on during the automatic calibration step. When you create a new experiment and you are at the window where you are inputting all of your experimental information, there is a measurement settings section. Under that section, the lasers you have available to you will be shown with boxes next to them. Make sure to select any lasers you need for your experiment here. No matter what, you always need the 488nm (blue) laser enabled for forward and side scatter (or back scatter as Sony likes to call it). Make sure for FSC and BSC that height, area and width are all enabled. If you need any more help, feel free to reach out. Thanks! - Kathy

Hi Jennifer, Thanks for the question.The questions I would have for you is which cytometer have you used to acquire data and which version of the Flow Cytometry Standard is your data? Most cytometers will produce data files in the FCS3.0 format where time is added as a keyword and then should be available to you in the analysis software as the final parameter in the list.

​@@DerekDaviesCytometry Thanks so much for your reply. My data is from MACSQuant Analyzer 16 and I'm using FlowJo V 10.8.1. The FCS files are version 3 and when I inspect a sample it does show the boxes for median parameters vs. time but there is not time option to actually open it up on x or y axis to cut out points where I know the machine clogged.

@@JenniferGrinder Hi Jenner! Would you mind sending an example fcs file to [email protected]? We can take a look for you and see if we can find out a solution. Thanks! - Kathy

@@OpenFlowCytometry I appreciate all your help! I figured it out, "time" in MACSQuant language is HDR-T, so we are all good. :)

@@JenniferGrinder Wonderful, happy to hear! A good start to 2023!

Hi Syeda. I'm not sure I understand your question. Did we state there were positive and negative autofluorescence controls somewhere in the video?

Hi Asisipo. We covered voltage optimization in another OpenFlow class: kzfaq.info/get/bejne/jM6ljaeBu56ZeIk.html. Check that video out and let us know if you have any questions. :)

@@RuiGardner Good day Dr. Gardner Thank you for your response and the video recommendation. The video has been very helpful when trying to set up the voltages for my panel, however, I got a bit confused, my panel has 12 colours, so I was wondering if I need one tube (cells and one fluorochrome) for each colour or if I can use one tube with all the colours mixed to set up the voltages? Another confusion was whether I need a specific number of cells (PBMCs) for each voltage run or if I just need enough for about 10 000 events? Thank you.

Thanks for watching Woo! At this time I don’t believe we can incorporate this. If you need any specific info don’t hesitate to reach out. Thanks! Kathy

Hi Matthew, Flow always gets a bit trickier when you get away from the small round cell paradigm! Vero cells will be larger and more epithelial than lymphocytes so the voltages used for FSC and SSC will likely be lower than if you are using lymphs and much lower than those set by any QC beads. If you are using a FSC v SSC dotplot be careful to see if there are any events piled up on the right or upper axes. You can bring these on scale by lowering the appropriate voltage. It is liley that there will also be much more of a spread of the cell 'cloud' than with beads or lymphs. If you are using a BD machine, you may be able to alter the ASF (Area Scaling Factor) - plot FSC-A v FSC-H - the cell population should have about the same value on each axis - if they do not you can alter the ASF to align them. Hope that helps!

@@DerekDaviesCytometry thanks so much! Super helpful!

Hi Marta! This is a great idea. Maybe we can plan another sorting class with a few specific applications like single-cell sorting. We'll brainstorm this. :)

Hi Ping! The negative does not matter here. It matters that the positive fully stained experimental sample is on scale and then we want to run the compensation controls at the same voltage. We go into a lot more detail on optimizing voltages in another session!

Dear Runhuan, thank you so much for your suggestion. We are actually working on ways to place these videos in our own channel on video sites like Youku or Bilibili, so we can better reach the flow community in China. We would love to know your opinion. Would you mind sending your contact information so we contact you. If you prefer, you can drop me, Kathy or Derek, a private message on LinkedIn with your email. Thanks again for contacting us with this suggestion. 🙂

Hi Carlos! Great question. This is something that may be easy to identify as debris or background as a cytometrist but not so easy if you’re not used to looking at this data all the time. We know in this dataset there is one homogenous population of cells by scatter we expect. This population is clearly identified and pulled out from the population that is very low on FSC and SSC which we typically identify as debris. What can be really beneficial if you’re unsure is to start off by looking at your fluorescent population of interest and gating on the positive cells (if there are any) and from there looking at a FSC/SSC plot to see where these events fall to help you set a gate for scatter. Another option would be to include a live cell marker such as calcein. Some messy tissue preps have lots of debris (think brain!!) and we can pull out metabolically active live cells from debris by using calcein, as it will be cleaved by esterases and fluoresce in live cells but not debris. Another option for you would be to run a sample that was buffer alone. You will see a lot of this debris and background there. Since you know there are no cells in this sample, you would expect all cells to be higher in scatter and can use this buffer alone control as a guide until you become used to scatter gating. I hope this all helps! - Kathy

@@kathydaniels1077 Thank you very much for your answer!! Everything's clear now! Amazing tutorial!

Hi Fabian, Diva produces .fcs files but they are stored in the database with numerical file names - to preserve the naming that was performed in Diva they need to be exported via the export function. It is also possible to export the experiment which would contain plots and gating etc. Sorting question: Doublet discrimination is never perfect and a doublet of a positive and negative cell could be sorted as a positive and this will impact on purity. This would not be the case if we sorted two positives or two negatives together.

Thank you JB C!

When mixing beads and cells, you do not necessarily need to run a separate unstained bead control. There is an option to have this unstained separate bead control in the software. If you do this, ensure that you specify in the setup that the negative reference for those samples is the unstained beads. If you do not do this, there is a negative/positive histogram gate in the unmixing setup that you can ensure has the correct gating. For this, make sure that your beads do have an internal negative control. It's also important to note that no matter what you will always need unstained cells to carry out unmixing. If you are working with multiple cell types with differing autofluorescence patterns, you will need to run an unstained for each of these and carry out group specific unmixing in SpectroFlo. Please let us know if you have any questions. Thanks! - Kathy

@@kathydaniels1077 Appreciate your detailed explanation! -Fei

Hello Fei Tu. Thank you so much for a great comment. It is always best practice to run your single color controls and calculate unmixing at each run. There can be differences in instrument/reagent performance from day to day and the only way to account for this is through running single color controls. Thank you! - Kathy

@@kathydaniels1077 I see! Thanks for your reply.

If you are using a BD machine and use the CS&T settings this is a good starting point especially if you are using lymphocytes but not everybody does! Sometimes the settings that CS&T gives are not appropriate for other cell types and that's where the voltration can be useful. I also find that with some cells, particularly when using far-red emitting fluorochromes that we can sometimes get a better separation by changing the voltages from those suggested by CS&T. You are right that we should make sure we are in the linear range of each detector and this is also a figure given by the CS&T program. In practice, I try not to have any signal above 10^5 when using Diva.

Hi Hadee, thanks for your question, it's a good one. Very often we don't see a binary positive a negative population in our samples but rather a shift above the control and often this is not particularly large. But we should be careful about normalising by division of a negative because we aren't comparing the same things. The positive signal will come from expression of a marker or fluorescent protein or probe but the negative signal will be autofluorescence primarily plus some background staining. There are some times when we do want to do this though; for example if we're looking at modulation of a level of a marker with some sort of treatment and then it is perfectly fine to calculate a ratio (or fold-increase/decrease) because we are then comparing apples to apples. When we have overlap of the negative and positive signals, so when our positive signal is weak ,it is sometimes better to take into account both the positive and negative samples and in these cases this is where the resolution distance metric can be useful. It is very similar to the staining index that we discussed in previous webinars and takes into account the difference between the median of the positive and negative populations but also uses the spread of the data (as measured by the standard deviation) of both of those populations. It is calculated as (MFI+ - MFI-)/Sqrt (SD-^2+ SD+^2). The bigger the number the greater difference between the two histograms. This is often used in imaging or imaging flow cytometry data. Hope that makes some sense!

Hi Stamatia, I would normally recommend starting by optimising the voltage for the channel by using the lowest dilution - this should give the brightest positive signal (slight caveat, check that there is not too much background staining by running an unstained control). At that voltage then run your titration samples and I normally derive a Stain Index for each point which will allow the optimal titre to be established. The reason we plotted BV421 against PE is that sometimes the positive population is easier to see in a dot-pot than a histogram especially when the signal is low and/or the cells are highly autofluorescent; we don't always have to use a histogram and cam derive the stats e need for the stain index from either a histogram or bivariate plot.

@@DerekDaviesCytometry Dear Dr Davies, thanks very much for coming back to my question. I will have to try this and see how it does actually work in my hands. there are still things I struggle though: shall I adjust the brightest signal at 10^5 by changing the voltage, in other words to get it on scale? And then, if I have 7 or 8 dilutions of my antibody shall I go through all this process to define the right concentration on the correct voltage (Voltage walk from 200 - 700V)? It is very confusing. I have a 17 colour panel and I cannot see how dealing with that.....

@@matini-hd Hi Stamatia! When starting out, if you are doing an antibody titration you can also load your sample that has the highest concentration of antibody. Make sure that the signal for the positive is on scale by adjusting the voltage. If in Diva, the first tick mark after 10^5 is typically where I want my brightest signal to be. Then you can be confident that all of your antibody titrations will be on scale. Run all samples at this voltage and then calculate your optimal antibody titer. From there, you can do a voltage walk to determine the optimal voltage to get best separation with the antibody concentration you are using. Does this help?

This is coming from Kathy, by the way!

@@OpenFlowCytometry Hi Kathy, you actually made my day much nicer! Yes it helps a lot and makes total sense now. I m on it next week and fingers crossed it will work out the way I want! Many thanks to you and the team for this great videos and supplementary explanations! 😊 Big hug from Heidelberg-Germany!

You are very welcome Sharareh!

Thank you so much Nathan! We are happy you're enjoying it!

Thank you so much for the kind words! We are glad you enjoy. - Kathy

Hi Akriti. There's a lot to unpack in your question. 1) In conventional flow cytometry there's a one-to-one relationship between fluor and channel, i.e., you pick the best channel that will detect a given fluor. Therefore, you don't have to worry about the voltages/gains in other channels. As we mentioned in a few of our sessions, first you optimize the voltages and gains of each channel in the cytometer. Then you titrate all your ab-fluor conjugates, and then you acquire every control and sample. The only reason to adjust voltages AFTER optimizing the voltages of each channel is when your samples or controls show signals of scale. Apart from that, you should leave ALL voltages/gains untouched regardless of spillover. Yes, even if the signal of a fluor is higher in a secondary (or spillover) channel than in the primary channel (the one you chose to detect that fluor). 2) Flow cytometry measurements are relative and displayed in arbitrary units. Since we can change voltages or gains, a positive signal can be anywhere in the scale. So signals above 10^3 can be either positive or negative, just like signals below 10^3 can be positive or negative. You must run a negative control to define what is positive and negative, and those signals can be anywhere in the scale. 3) I'm not sure I understood your final question, but if the antibodies were not well titrated, or your instrument was not well optimized (we call it "instrument characterization", which means optimizing your voltages/gains for each channel) you may not see a good separation between positive and negative populations. So the problem you seem to be facing (again, apologies if I'm not understanding your question fully) could be related to the instrument not being well optimized in the first place, and/or insufficient titration. Did this answer your question? Please feel free to elaborate and we'll be happy to answer.

@@RuiGardner Thank you for the quick reply! Yes, you have partly answered my query. But I still did not understand most of it (sorry!) I shall delineate as to what I do - 1. I prepare single stained controls, FMOs and unstained control using splenocytes while the samples that are stained fully would be meninges 2. At the cytometer (we use FACS CantoII here), this was what I was taught. i) First, to adjust/optimize my voltages, I run the unstained at low and check if the peaks of all the other colours are below 10^3 (without recording). ii)Then, I run every single control individually (without recording ofc) to set the voltage in order to ensure that the single stain control in question is beyond 10^3 (or brighter for positively stained) and for rest of the colours at the same time, their peaks need to be less than 10^3 log scale. I consume alot of time here because certain colours like PCP have spill overs into APC, PE Cy7 and even AF700 channel on Canto II. Therefore, I ended up spending a lot of time in reducing APC's, PECy7's and AF700's voltages only to ensure that when I add PCP single stained tube at SIP, I must get (idealy, I think) a positive peak beyond 10^3 with the adjust voltages! what I also came across was that therenwere small peaks in the other 3 channels were beyond 10^3 (just like in PCP's channel) while running PCP single stained tube (and I ensured that the single stained wells weren't conatminated by other colours!) Am I doing it right? Also, I could not follow up on the difference between adjust and optimizing the voltage that you have stated above in the answer. Please pardon me if I am asking a silly question!

@@akritino7437 Agreed with all that Rui said! Also, not a silly question by you at all! We often see that researchers are trained in this way based on older ways of thinking for instrument setup. The way that you are explaining is only going to cause you confusion and headaches. One of the best approaches for setting voltages is to run your single stain controls at increasing voltages. Since you are working with the splenocytes, I imagine that you have a positive and a negative population for most all of your single color controls. You record files at increasing voltages until you see that the signal is off scale (Just beyond 10^5). You can then calculate the Stain Index for all samples to determine the voltage that gives you the optimal separation of positive and negative. We go over this in depth here: kzfaq.info/get/bejne/jM6ljaeBu56ZeIk.html Now as far as decreasing other detector voltages in order to minimize spillover, this is not necessary! Remember, you want to optimize to get the best separation. If you start decreasing voltages, you will reduce the separation between positive and negative and risk the potential of not seeing dim or lowly expressing markers. Some people are worried about compensation values over 100%. That is misguided. If you look at this Flow Post-It by Barbara Oliveira you can see that adjusting voltages to lower compensation does NOT mean you get a better separation of signal. wi.mit.edu/sites/default/files/2021-12/20211004%20SOV%20over%20100_0.pdf I suggest you watch the voltage optimization OpenFlow and follow the guidance there to set your voltages. Though I will add that if your experimental sample is a different cell type than the sample you are working with for your single color controls, there may be a large difference in autofluorescence. If your experimental sample has higher autofluorescence, this can impact the optimal voltages, and you may need to reassess with the cell type of interest, taking into account the impact that difference in autofluorescence has on the separation. Best of luck, and please let us know if this all makes sense! Kathy

@@RuiGardner Okay, so I have seen your videos just to get my doubts further clarified. So, in a nutshell, just to be sure that I have gotten the concept right are the following - 1)For instance, if a voltage set is appropriate for particular colour, say PE, we must choose that voltage inspite of the fact that other colours could interfere with their signals popping up beyond 10^3 at that set volatge. Am I right? 2) So, the negative of any colour is based on the unstained population. So depending upon where our unstained population has been set, one doesn't have to bother about the decades on the log scale as long as they (unstained cells' curve) remain on the left hand side of the graph. Am I right again? 3. I used to previously assume that voltages could correct the compensation values. i.e. If I choose a voltage for a particular colour, A, at say 450, that voltage must be adjusted in such a way that the interfering signals from other colours at that voltage must not exist beyond certain voltage (10^3 since below this log decade, we had set the voltage of unstained). This adjustment would consequently reduce the compensation. That was what I had previously thought of. Therefore, it could be right to say that adjusting voltage does not necessarily reduce compensation. Rather, it simply changes the values of compensation. Am I correct over here? Last question. I agree that the autofluorescence of single stained control cells, unstained cells and our sample cells need to be the same. However, if one is using splenocytes to optimize voltages and compensate only and other cell types as samples, one could slightly change the scatter voltages of the sample while running. Is this correct? Or is there is any limitation to this? A point that I understood (and what my mind extrapolated from what Dereck stated in the video), was that if an unstained control is replete with Mac/DCs like in one's sample (even if the two belong to different organs), that is still fine as long as te autofluorescence remains more or less the same. Would this be the limitation? Am I making sense? That 's all I have got so far ! Please correct me if I have erred anywhere here! I just want to get my concepts right and be confident!

@@akritino7437 Hi Akriti, Many thanks for your great questions and hopefully our answers and other videos have helped you. But in answer to your questions: 1. It is very important - in fact perhaps the most important part of your experiment - to set the correct voltage for any particular channel. We then deal with overlap by using compensation but the value of compensation shouldn't be a primary concern. If you find that one colour is brighter in another channel than that channels fluorochrome then it may mean that this is not a good combination of fluorochromes. 2. As long as the positive and negative are as far apart as possible and the positive population is not above 10^5 then it doesn't really matter where the negative population is. We should get away from the idea that our negative population should be right in the bottom left corner of the plot. 3. Changing voltages will change the compensation values But I would caution against doing this because, particularly if you are reducing the voltage, you will be lowering sensitivity and potentially missing weakly positive cells. No matter what the value for compensation though the spread of data ,which is the thing that will reduce resolution in the spillover channels ,will remain the same. 4. It is perfectly acceptable to change the scatter voltages during running these do not affect compensation. The concept of ensuring you compensate to an unstained autofluorescent population becomes important in heterogeneous cell populations. If you had a mixture of say monocytes and lymphocytes and you compensate your monocyte marker to the unstained monocytes level of autofluorescence and lymphocyte marker to the unstained lymphocyte level of autofluorescence then the compensation will be equally applicable to each cell type as long as you haven't changed any voltages! Let us know if you still have any outstanding questions.