FAQ  •  Register  •  Login

Acceptable / typical background levels

<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Mon Apr 29, 2019 1:45 pm

Acceptable / typical background levels

Hi all,

Just for fun (and since I'm still learning the instrument), I ran a solution of 18.2 M Ohm dH20 with EVERY channel / parameter enabled to see what it looked like. The results are below:

Image

Those are dual counts in solution mode.

Things I've learned from this (and other reading / learning):

1) The Ar dimers are a natural "byproduct" of the plasma.
2) We always seem to have some iodine and lead in our system.
3) Xe is present in Argon tanks and tends to increase towards the end of a tank.
4) The instrument cannot distinguish between isotopes of the same mass.
5) The method to make a best guess of which isotope is responsible for a signal is to look at the abundance of both the isotope and the element.

For example, the Ba/Xe 132 signal is most likely Xe because:

a) There are other "pure" signals from Xe.
b) 132 is one of the common isotopes of Xe.
c) Ba mostly exists as 138 and a 132 signal would therefore only be a result of Ba if also accompanied by a VERY high Ba 138 signal.
d) The La/Ce/Ba 138 signal is almost certainly Ba because it's the most common isotope for Ba. Ba is also more common on earth than La and Ce (in addition to La and Ce being very rare in 138 form).

So I guess my questions are:

1) Given this is the "cleanest" signal I'm likely to see, what's an acceptable level of background from other contaminants when running samples?
2) If there is a high level of some isotope, but it's not enabled, is it still contributing to detector aging?
3) How much of a concern is detector aging?
<<

mleipold

Guru

Posts: 1778

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Mon Apr 29, 2019 4:04 pm

Re: Acceptable / typical background levels

Hi James,

Generally, I agree with your conclusions.

1. Yes, Ar dimer is something inherent to the instrument. There are 1-2 settings that your Fluidigm engineer can adjust to (slightly) alter your Ar dimer signal, but you cannot get rid of it. Frankly, I find it useful to have as positive signal: if Ar dimer goes away, there are significant issues with the instrument's overall signal/sensitivity.

2. Iodine, Barium, and Lead are probably the 3 most common contaminants, in my experience. They often come from buffers used in staining or cell isolation, or from the water used. 3% nitric acid (or Tuning Solution, which contains 2% nitric acid) does the best job of stripping them out of the system.

3. Yes, Xe is always present in Ar: they're both noble gases present in air. Argon boils at 87K, while Xe boils at at 165K. Xe is present at a much lower concentration in air than Ar (http://scifun.chem.wisc.edu/chemweek/GasesofAir2017.pdf), so it's probably not economical to separate it out unless there's a *specific* reason to. Liquid Xe is also more dense than liquid Ar, so it settles in the liquid dewar and does generally increase toward the end ("heel") of the liquid tank. I haven't seen much, if any, variation in Xe signal over the course of a compressed gas cylinder.

Again, I like seeing the Xe: i the Xe goes away, that usually means that there's a problem with the instrument (usually liquid sample flow problem; for some reason, the Xe signal in wet plasma is higher than the Xe signal in dry plasma).

The main issue with Xe is when you get a "bad" tank where the Xe levels are too high. The machine uses Cs133 in the Tuning solution in the Mass Calibration process: it has a small TOF "search window" around the expected Cs133 arrival time, and mainly locks onto the brightest signal in that window. If the Xe132 or Xe134 signals are too high, it has trouble, and ideally refuses to calibrate (so Tuning would fail at the Mass Calibration step). The software has a default max Xe signal for this reason.

4. Correct, the machine doesn't know what an element is....it only knows that ions are arriving in a time window that you have calibrated to mean a certain mass signal. So, it can't tell the difference between Cd110 from a Qdot and Pd110 from a BC agent, or from Gd155 vs La-oxide.

5. Generally yes: however, don't forget oxide spillovers.



Detector aging does happen. This is one reason why you don't want to run really streaky samples, as the streak ions are still hitting the detector without you getting "useful" data. Also, the more channels you use and the more metal signal in each channel, the more ions hitting the detector and the faster it ages. This is definitely one reason why you want to titer all your reagents (including BC agents, Live-dead, and Ir intercalator).

Yes, even if you're not measuring all possible channels, all ions present in the sample in the measurable mass range hit the detector, and therefore contribute to detector aging. This is the difference between watching your sample in Masses view vs TOF view. This is also why I generally watch my samples in TOF view, to be able to detect background signals in other channels *not* being collected. First, because I just want to know if there are other contaminants. Second, occasionally those "absent" channels contribute to signal in "measured" channels.

For example, I don't normally have a probe on La139, so I don't normally collect that channel. However, a couple years ago, I had a sample that was (unknown to me or my customer) hideously contaminated with La139. The La139 signal was so bad that it looked like a solid black Sharpie line in TOF view. The La138 signal was also almost a complete line, and even though there's no La140 isotope, the right-leg spillover from the massive La139 peak was noticeable in the "Ce140" bead channel. Additionally, the Gd155 channel was super-streaky as well, due to the massive La-oxide spillover.

Since I had a weak marker like CCR6 in that channel, I basically couldn't interpret that channel in that sample.


Mike
<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Tue Apr 30, 2019 2:30 pm

Re: Acceptable / typical background levels

Thanks Mike.

Very useful as always!
<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Wed May 01, 2019 4:27 pm

Re: Acceptable / typical background levels

I’ve realised I’d still like a guideline on what level of contamination is acceptable.

ie when looking at the TOF plot, at what point do you tell the user their sample is too dirty and to try methods to reduce contamination?
<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Tue May 07, 2019 2:19 pm

Re: Acceptable / typical background levels

As an example, I've compared the 15M water vs. 18 M water in our lab.

In solution mode (dual counts), I see that they are almost identical:

Highest contaminant common to both is I 127 (~18k DCs).

However, the 15 M water also has Ba 138 at around 28k DCs, whereas the 18 M it's almost not present @ 95 DCs.

How do I decide if the 15 M water is "good enough" (for washing etc.)

Do I run it in solution mode to quantify the DCs, or do I look at it in the Rain Plot in TOF mode and qualify it by looking for "tyre tracks" or black bands?

The same question would also apply to PBS. I've compared three (our DIY version, some Gibco stuff and the Maxpar).

The MaxPar (in solution mode) shows Ba 138 about 16k DCs and Ce 142 around the same.

Our DIY PBS has Ba 138 up around 62k and Ce 140 around 58k.

The Gibco is pretty clean (nothing above 8k DCs), *except* for Rh 103, which is around 45k DCs. If it helps, that looks like this on a rain plot (intensity):

Image
<<

mleipold

Guru

Posts: 1778

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue May 07, 2019 2:55 pm

Re: Acceptable / typical background levels

Hi James,

Ultimately, you want only desired signal, and no undesired signal (plus things you can't remove, like Xe and Ar dimer).

I'd advise using only 18MOhm water, unless there's some really pressing reason why you're looking at 15MOhm water. I would even use 18MOhm for washing, was well as buffer making (10X PBS down to 1x PBS, etc) and sample dilution. Part of it is that in my experience, anything less than 18MOhm tends to be variable/unreliable: you listed some values from your 15MOhm today, but it might be different (or a different ion distribution to give the same resistance) tomorrow.

The Gibco PBS doesn't look like anything I'd want to use. Again, historically, some of my users who insist on using it see lot to lot variations in contaminants (both type and intensity). I think your 103Rh streak in the Masses screenshot is high enough that I wouldn't want to have it running constantly into the sample.

Additionally, there's some contaminant at or a bit above your 7939 TOF. Since you're not seeing a streak in the 90Zr channel, it can't be Zr. Looks like it's Kr (84Kr ~50% nat abund) or Sr (88Kr ~83% nat abund)? If it's Kr, then it's presumably coming from your Ar supply. If it's Sr, then presumably it's coming from your Gibco PBS for some reason.


In either case, I still wouldn't want it running in my instruments.


Mike
<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Wed May 08, 2019 7:54 am

Re: Acceptable / typical background levels

Thanks Mike - really appreciate it!

Just for fun, I loaded the image into a really useful online tool I used in a previous professional life - https://automeris.io/WebPlotDigitizer/

This enables the extraction of XY values from images.

Using that (and a spreadsheet I made to convert TOF -> Mass and vice versa), the most likely contaminants in the Gibco are:

7755 = 84 Kr (although 18 M water does not show this, so perhaps it's not from the Ar supply?)
7800 = 85 Rb
7980 = 89 Y
8773 = 108 Pd
10296 = 149 Sm

18M water for comparison:

Image
<<

mleipold

Guru

Posts: 1778

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Wed May 08, 2019 2:58 pm

Re: Acceptable / typical background levels

Hi James,

Yes, your 18MOhm water looks like I would expect: nothing really but Ar dimer and Xe isotopes.....the trace iodine you have is probably remaining from a sample, and would strip out if you used 3% nitric (or the 2% nitric in Tuning Solution).

If you're curious about TOF/mass conversion (or, more precisely, have some streaks you can't quite identify), you can look at the TOF Plot in the Helios software, left side (Plot/Rain plot, TOF, MPR, Plotviewer). You can toggle back and forth between TOF and Mass, looking at the spikes in the plot. You can move the mouse crosshairs over to find out what mass a given spike corresponds to.


Mike
<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Wed May 15, 2019 1:27 pm

Re: Acceptable / typical background levels

Thanks Mike.

In that case, I think I can answer the question quantitatively:

Given that:

1) Maxpar PBS "worst" contaminant is Ba 138 and is giving 16,000 max DCs in solution mode.
2) You say that you don't like the look of the Rh 103 in the Gibco PBS on the rain plot, which corresponds to about 40k DCs in solution mode.
3) You also say you're not keen on 15M water, which gave me 28k DCs in solution mode for Ba 138.

I propose that <20k DCs in solution mode should be sufficiently "clean".

(I much prefer to quantify things where possible, rather than judging a rain plot qualitatively, although I realise that whilst running, one tends to be in event mode and therefore the rain plot is all that can be shown, therefore it's important to be able to recognise that too).
<<

jimbomahoney

Master

Posts: 60

Joined: Wed Feb 27, 2019 11:21 am

Post Thu May 16, 2019 12:34 pm

Re: Acceptable / typical background levels

This figure of 20k in solution mode also correlates with an e-mail exchange I just had with Fluidigm.

I'll paraphrase / summarise (again, correct me if I'm wrong):

  • In tuning solution, the Ir 193 has the lowest signal - about 400k in Tuning mode (100k in solution mode because this is only a 1 second accumlation, rather than the 4 sec in tuning mode).
  • The ratio of lowest signal:highest contaminant should be >5.
  • Therefore, the highest acceptable contaminant would be ~100k / 5 = 20k (in solution mode).
Next

Return to Introduction to mass cytometry

Who is online

Users browsing this forum: No registered users and 1 guest