FAQ  •  Register  •  Login

Single Alive cell counts for analysis

<<

jobagins

Participant

Posts: 13

Joined: Thu Oct 22, 2020 6:45 pm

Post Mon Aug 30, 2021 3:21 pm

Single Alive cell counts for analysis

Hi fellow mass cytometry friends.
I wonder about your opinion on good cell counts to analyze mass cytometry data. Lets imagine any regular panel staining of human PBMCs, surface and intracellular, with transcription factor markers.
What is the Single Alive count that makes you comfortable to use for any downstream analyses?
I love to see +200K but know its not always possible to get this yield, especially from scarce clinical samples.
Looking forward to hear you on this one, where do we draw the line
joanna
<<

mleipold

Guru

Posts: 5792

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Mon Aug 30, 2021 3:37 pm

Re: Single Alive cell counts for analysis

Hi Joanna,

I think there are two things in your one question:
1. Viability (and to a lesser extent, singlets).
2. Cell count.

1. Viability: This is also in 2 parts: how long it takes you to acquire however many LiveIntactSinglets (poorer quality sample = longer acquisition), and the impact of poor sample quality on proper staining (somewhere around 60% viability, the staining gets *noticeably* worse even in the LiveIntactSinglets).

2. Cell count: I think it comes down to what your question is, and the rarity of your cell population. In another thread (viewtopic.php?f=6&t=1980&p=5160) , Cynthia Guidos talked about and gave a calculator for counting statistics ("You need to decide on how precise you want your measurement to be and to estimate the frequency of the rarest subset that you want to measure").

So, if you're looking at a phospho panel (eg, pSTAT1), in most cases an entire parent population such as CD4+ will respond to the stim. Put differently, the entire CD4+ pSTAT1 histogram will shift to a higher signal. Since the entire population is responding, you have more events to satisfy your Cell Count.

In contrast, for ICS, maybe only *some* of your CD4+ will make a particular cytokine upon stim. So, you'd need proportionately more CD4+ (and therefore total Cell Count) in order to satisfy Cell Count for IFNg+ CD4+.


In an extreme case like antigen-specific T cells (peptide stim, or tetramer), you may need to collect a *lot*.


In a good viability PBMC sample, a rule of thumb at the Stanford HIMC we aim for:
1. surface pheno (no ICS): 250K Ungated/sample -> ~150K LiveIntactSinglet.
2. ICS (surface + cytokine): ~300K Ungated/sample -> ~200K LiveIntactSinglet.
3. Phospho (surface + phospho transcription factors): ~150K Ungated/sample -> ~100K LiveIntactSinglet.


Mike
<<

cguidos

Contributor

Posts: 28

Joined: Tue Nov 18, 2014 3:10 am

Post Mon Aug 30, 2021 3:38 pm

Re: Single Alive cell counts for analysis

Hi Joanna - to answer this question you need to have an idea about the rarest subset you want to measure, and the precision with which you want to measure it. I have tried to upload an attachment that takes you through the calculations for different subset frequencies. Hope this helps!

Cynthia
Attachments
Counting Statistics and the Poisson distribution 20200729.pdf
(234.17 KiB) Downloaded 200 times
<<

cguidos

Contributor

Posts: 28

Joined: Tue Nov 18, 2014 3:10 am

Post Mon Aug 30, 2021 3:41 pm

Re: Single Alive cell counts for analysis

Hi Joanna - looks like Mike partly covered my answer in his reply. I have pasted some key metrics here in case you can't access the PDF I attached:

The total number of cells you should collect is determined by:
1) the frequency of the rarest population you want to measure and
2) the precision, represented as the coefficient of variation (CV), with which you want to measure it.
• The precision of your measurement is calculated using the Poisson distribution based on the number of events you count in your subset of interest.
• The essential feature of Poisson distributions is that if N events are observed in your target subset, then the standard deviation (SD) associated with that count is square root of N.
• The coefficient of variation (CV) is then given by: CV(%) = 100/sqrt N.
• Thus, as you count more cells in your target subset, the CV will decrease, indicating higher precision.

For most applications CVs in the 2-5% range are considered ideal:
• For a CV of 5%, you need to count 400 cells of your population of interest.
• For a 1% population, you will need to count 40,000 total events to achieve a CV of 5%.
• For a 0.1% population, you will need to count 400,000 total events to achieve a CV of 5%.
• Note that if your sample has many dead cells and/or doublets you need to consider the % of total cells collected represented by your target, not the % of live singlets.
• Lower precision (10-20%) is generally acceptable for rare event analysis when it is not feasible to collect enough events to achieve higher precision
<<

jobagins

Participant

Posts: 13

Joined: Thu Oct 22, 2020 6:45 pm

Post Tue Aug 31, 2021 2:55 pm

Re: Single Alive cell counts for analysis

Thank you Cynthia and Mike, this is great. This totally align with our procedures. Agree with Mike that below 60-65% of viability the staining is getting "flaky".


Thank you for your insight!

Return to CyTOF general discussion

Who is online

Users browsing this forum: No registered users and 10 guests

cron