Barcoding combined with MDIPA kit II

[DanielBachurski]

Dear all,

I am choosing a live cell barcoding approach to analyze PBMC subpopulations in leukemia patients in combination with the Maxpar Direct Immune Profiling Assay. I have a few questions concerning the measurement capacity of a Helios device and look forward to your suggestions and comments:

- What is the maximal amount of cells in a maximal volume that can be measured in compliance with the recommended flow rate of a Helios device in a 5 ml polypropylene FACS tube? What are your recommendations for cell number, volume, and flow rate for a 20 sample 6-choose-3 approach in one tube if I aim to maximize the cell number? - Leukemic cells reduce the percentage of PBMCs in the sample, which makes it sometimes tricky to analyze scarce subpopulation.

-What is the maximal cell number I can stain in one tube of the MDIPA titrated antibodies (Fluidigm) in a certain volume? Did some of you titrate the MDIPA kit for this purpose in combination with live cell barcoding? Also, any recommendations here regarding maximal cell number, and staining volume?

Thank you for your help and best regards!
Daniel

[mleipold]

Hi Daniel,

I'll address only some of your questions....for example, I haven't really used the MDIPA tubes much yet.

The PSI unit on a Helios has a maximum tube volume size of 5mL. We typically do not run samples more concentrated than 0.8M/mL (as measured by a Biorad TC20; corresponds to about 250 events/sec), so that would give you a maximum of 4M cells in that 5mL.

With a 40% recovery (cell transmission efficiency), that would give you about 1.6M theoretical cells. This would be total cells: in your case, leukemic and non-leukemic. I can't give you any guidance on fraction of non-leukemic cells in a leukemic sample.


The flow rate of 30uL/min is essentially fixed on a Helios PSI.


Mike

[DanielBachurski]

Hi Mike,

Thank you a lot for your helpful response!

I thought that the staining capacity of the MDIPA kit would limit my PBMC barcoding experiments, but I do not worry about that anymore, unfortunately!

Maybe someone frequently analyzing barcoded PBMCs can report his experience with Helios and the number of pooled samples and total cells that still allow for in-depth profiling of PBMC subpopulation.

I guess it is not possible to improve the 40% cell transmission efficiency, right?

Best
Daniel

[mleipold]

Hi Daniel,

There's no real way to increase the 40% cell transmission efficiency. I mean, 40% is a bit of a lower bound, but realistically with the PSI or with the Supersampler, you're at best going to get 50%. So, better to do your calculations with 40% and err on the side of caution.

I'm not sure I understand your other point about PBMC MDIPA+Barcoding. According to this, MDIPA is designed to stain 3M PBMCs or 270uL WB in one tube:
https://www.fluidigm.com/binaries/conte ... igm%3Afile

I'm sure there's a little wiggle room there, but I wouldn't advise staining 500K or 10M in just one MDIPA tube, you'd be over or under-titered. If you have live-cell barcoded as a first step, then combined into a single sample, you would want to *then split* the combined BC sample into aliquots and stain in separate MDIPA tubes of 3M. Or, conversely, dissolve the MDIPA lyo reagents from however many tubes you need to use, combine to a single cocktail tube, and then use that to stain your single combined BC sample.

I strongly suggest that you discuss your assay plans/needs with your Fluidigm FAS: they may have internal info on the lower and upper bounds of cell number that you can stain per single MDIPA tube.


Additionally: when you barcode, each sample affects the other samples. This means that if you have 19 good samples and 1 really poor-quality sample, any clumping and/or streaking from that poor-quality sample affects all 20 samples, 19 good and 1 bad. This also means that if you have strongly varying sample types in a given combined BC sample, each sample's cell distribution will affect the acquisition distribution of the entire composite sample. As an extreme case, imagine that you had a B cell line sample and a T cell line sample. The combined sample would be 50/50 B/T, and you'd have to take that into account for your target number of acquired Ungated events. So, you'd have to acquire 250K events to get 125K T cell events.

You may have previous data from flow experiments that will inform you on likely cell distributions/frequencies in all your planned patient types. Use that as a way to "guess" how many cells in total you'll need to acquire from the composite BC sample in order to still reach the depth of profiling you want to achieve in *each* donor sample.


Mike

[cguidos]

Counting Statistics and the Poisson distribution 20180108.docx

(21.73 KiB) Downloaded 741 times

Counting Statistics and the Poisson distribution 20180108.docx

(21.73 KiB) Downloaded 741 times

Counting Statistics and the Poisson distribution 20180108.docx

(21.73 KiB) Downloaded 741 times

Hi Daniel

The number of events to collect should be guided by 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. I have attached an explanation of these concepts and an Excel calculator that we use to plan run times per sample. You'll see that what matters is the total number of events in your final gated population. We typically aim to collect 100,000 events per PBMC sample. With this total we can measure about 1,000 events for a 1% population, which gives us a %CV of 3% which is excellent. For a 0.1% population we are only counting about 100 events, and the precision drops to about 10%, but this is acceptable for rare event analysis. For barcoded experiments if we have multiplexed 10 samples into a single tube, We would collect about 1 million total events.

Hope this helps!

Cynthia

[DanielBachurski]

Hi Mike,

You are entirely right: MDIPA offers a titrated staining for 3M cells. I will strictly adhere to this recommendation and use the lyophilized ABs of more staining tubes if the sum of my barcoded PBMCs exceeds 3M.

I intend to the barcode to reduce the variability of a multi-step intracellular staining. I will start with ten barcoded samples in one tube aiming to capture 100,000 cells of each donor as Cynthia suggested and trying to account for the fixation/permeabilization caused cell loss during washing steps.

Thank you!

Hi Cynthia,

Thank you a lot for your help! I will try out the barcoding of 10 patients accordingly and use your calculation sheet to evaluate my experiments!

Best wishes
Daniel

P.S.: This forum is fantastic and already helped me a lot in setting up my experiments! Thanks to everyone involved here!

[mleipold]

Hi Cynthia,

One question to clarify: you said "We typically aim to collect 100,000 events per PBMC sample. With this total we can measure about 1,000 events for a 1% population"

But to get 1000 events in the 1% population gate, that would be 100K LiveIntactSinglet events, right (ie, Debris, Dead Cells, Beads, and all other junk removed)?

If so, then assuming 40% cell transmission efficiency, you would need to actually acquire 100K/0.4 = 250K Ungated events hot off the machine, right? FYI, 250K Ungated events is what we typically acquire per PBMC phenotyping sample here at the Stanford HIMC.


Mike

[cguidos]

Hi Mike

If you look at my Excel caclulator you'll see that I included the efficiency calculation and calculated theoretical cell# exclusive of beads. For PBMC we typically get 70-80% efficiency calculated as the # events collected vs # sampled. The latter caluation is based on accurately measuring the cell concentration right before acquiring and the volume acquired based on the flow rate.

But you can plug in whatever efficiency value makes sense to you and the numbers will auto-update. So if efficiency is 70% we would acquire ~150,000 events to end up with around ~100,000 in the FCS file. The caveat of course is that the latter number includes dead cells but in our PBMC preps viability is usually >90% so I don't worry too much about that.

If you are collecting 250K events then you have fabulous CV's for 1% subset (3-5% is considered excellent) and 6.3% for a 0.1% subset - so better than 10% CV but still not below 5%. To get to 5% CV you need to collect at least 400 events in the final gated population, which would require 420K total events under my assumptions above. We have some clients who insist on collecting 500K events per sample which takes almost 1 hr. Mathematically that only slightly improves the CV for 0.1% subset and to me is not worth the extra time. My attitude is that if people want to accurately measure subsets that are <1% of total they should pre-enrich first!

Cynthia

[mleipold]

Hi Cynthia and Daniel,

Cynthia: sorry about getting my percentages mixed....you just beat me to correcting my post. I would agree, for a standard decent-quality sample, LiveIntactSinglets being 70% of Ungated Acquired events is very typical.

Daniel: based on those calculations of Cynthia, you'd want to acquire *150K Ungated* events per subsample in order to get ~100K LiveIntactSinglets per subsample that you would then go ahead and do the rest of your marker-based analysis on.


Mike

[Ezequiel]

Hello Daniel!, I intend to do something similar to your protocol but with Whole blood. I was wondering if you had gone ahead and tried the experiment. How did it go? I would really appreciate any advice.