



Facility News
Summer is upon us and according to the prevision, it will be a very hot ☀️ and stormy 🌩️season ! What a better way to protect yourself than to spend more time in our analysers room which is always in some sort of winter ❄️ level cold zone ? Always look at the bright side in every situation 😉

This month, in FACS Tips we are answering a recurring question in our facility : Is my population real ? We often have to discuss with users about it and we will provide here an explanation so you can be confident your data is revelant.

The Attune CytPix is finally here until July 8th so if you are curious to try the machine for FREE, please contact the FCF Team about it !

The May Newsletter Quiz has been won by Marco Ongaro from the Verdeil Lab, congratulations !



He correctly answered the first 3 questions (CD32, HBSS and True) and estimated FACS Flow usage for April at 717 L while the correct answer was 660 L ! Good job !

Do you want to win this tasty FCF Toblerone ? Please find the questions for this month Quiz at the end of the Newsletter !

Finally, I would like to remind everyone that the past FCF Newsletters are always available on the website at this location.



FACS Tips

How many events do I need to record?
When performing a flow cytometry experiment, often samples are just run to completion. However, much is there is how much we record. While this is often the safest approach to an experiment, it is unprecise and can come with some downsides. More often this will give us much more than we need, meaning we use an unnecessary amount of machine time, and slow ourselves down with larger than needed file sizes. It is also possible that we may be under sampling and need to scale up the number of cells in each tube to make more precise repeatable experiments. This decision of how many cells to record can be advised by looking at the Poisson distribution of our sample.

The Poisson distribution, in statistics, is a probability distribution that is used to predict how many times an event is likely to happen over a specified period of time when we are given the average number of times that event occurred in the time period. While this may sound intimidating it is actually quite simple to calculate.

To determine the number of events to be recorded in our target population for our experiment, we can just focus on the equation below. From this value we can also determine the estimated number of total events it will take to achieve this value.



n represents the number of cells needed in the target population to achieve the desired CV
CV Coefficient of Variance is the quotient between SD and mean and represents the precision/reproducibility of the experiment

Now the challenge is to determine what is your appropriate CV value to aim for. Because CV is a product to the standard deviation (SD) divided by the mean, as we increase the mean (number of events in our target population), our CV value goes down. This is important as it represents increased precision and reproducibility of our experiment. If our goal is to lower the CV, then that means we just need to increase the number of events in our final population of interest. So, if we start out with a chosen CV of say 5% (like in the example above), we know that we need to record 400 events in our population of interest. And if the frequency of the population is only 0.1% of the total events, then we need to record an estimated 400,000 events to get there.

It is suggested that the biological variation is around 20% CV, so it would be advisable to aim for a CV lower than that, but there is no definitive value you must meet. We suggest aiming for a CV at or lower than the typical biological variability of a biological experiment which is 5%. Below is a reference table for the number of cells in a target population needed to reach certain CV values, and how many total events would be needed depending on the frequency of that target population.



Alternatively, if you want to calculate the interval of confidence of your results based on the number of cells acquired and the percentage of your population of interest, you can use the excel tool provided herein.

A more abstract topic in this realm is to decide what events are “real” versus not, and how many cells do we need to have recorded to say that a population/effect is really happening. In his paper “How many events is enough? Are you positive?”, author Mario Roederer tackles this issue (Cytometry A 2008 May;73(5):3845. doi: 10.1002/cyto.a.20549) . The main takeaway is that there is no arbitrary number needed to say a population is real or not, and that one cell in a gate is enough to be considered real, as long as you can prove that every appropriate control has been performed to lead you to the gates that were drawn to include that one cell. By the same logic, even if you have 100,000 events in your target gate of a poorly controlled experiment than it’s possible that the effect is not real, but rather background.

If we know the frequency of our population of interest and have agreed on a degree of precision for our experiment (CV) we can start with a goal number of cells to record that should provide us quality data by using the Poisson Distribution. Again, it is almost always better to record more events, but if we can reduce our file sizes and time on machine we can improve efficiency. At the end though, experiments even with a very small number of recorded events in our target population can be considered real and successful, as long as we have proved this with appropriate controls. As always feel free to reach out to the FCF staff if you have any questions.




Can you answer this ?




