FAQ  •  Register  •  Login

Strange signal shift

Forum rules
Please be polite and civil. We know that troubleshooting is vexing...
<<

Chowduck

Contributor

Posts: 29

Joined: Wed Nov 19, 2014 4:39 pm

Post Wed Oct 19, 2016 6:32 pm

Re: Strange signal shift

Hi Johanna,

This is a very interesting problem, thank you for sharing. I think Mike’s comment about debris binding antibodies preferentially is the most probable scenario to fit this channel specific drift over time. I would speculate there are debris particles that are positive for one or more of your 4 drift channels that are aggregating onto your cells over the 1h run.

With your current data set, you could gate on the first 5 minutes of your 1h run when the signal is low, then on the last 5 minutes during the peak of the drift. Then compare the frequency of DNA- events that are also positive for the markers in your drifting channels for these two time windows. If this is being caused by debris aggregating to your cells I would expect to see the frequency of DNA-, drift channel+ (148Nd, 150Nd, 152Sm or 154Sm) events per 5 minute window decrease towards the end of the run as they adhere to DNA+ cells over time (assumes a constant sample introduction rate).

If you were to encounter this again I would suggested pausing the run, more aggressively mixing the cell suspension, then filtering (35 micron nitex) before resuming to see if that corrects the drift you're seeing.

Please keep us updated, I’m curious to know the cause.

-Greg
UHN/Sickkids FCF
Toronto, Canada
<<

Johanna

Participant

Posts: 8

Joined: Thu Jul 28, 2016 5:23 pm

Post Wed Oct 19, 2016 7:29 pm

Re: Strange signal shift


With your current data set, you could gate on the first 5 minutes of your 1h run when the signal is low, then on the last 5 minutes during the peak of the drift. Then compare the frequency of DNA- events that are also positive for the markers in your drifting channels for these two time windows. If this is being caused by debris aggregating to your cells I would expect to see the frequency of DNA-, drift channel+ (148Nd, 150Nd, 152Sm or 154Sm) events per 5 minute window decrease towards the end of the run as they adhere to DNA+ cells over time (assumes a constant sample introduction rate).


Hi Greg,
thanks for your suggestions...

I did as you proposed and did not see a decrease in DNA-/drift channel+ events. When I use a generous DNA- gate (gating on everything with Ir signal below that of intact cells), I even see an increase in these events. If I gate more strictly on events with no Ir signal, I only see an increase in DNA-Nd150+ events, with no apparent changes in the other channels.

I'm staining cells today to see if I observe something similar again, and I will update when I know more.

Johanna
<<

Chowduck

Contributor

Posts: 29

Joined: Wed Nov 19, 2014 4:39 pm

Post Wed Oct 19, 2016 8:30 pm

Re: Strange signal shift

Thanks for the update. I pitched this idea do my colleague, her response was: if there is debris positive for the drift markers they would likely not have enough of a signal to get above the lower convolution threshold to trigger a cell event. This would explain why my suggestion did not work.

From your observation that the DNA-/drift channel+ events increase over time and my colleagues comments, maybe as the debris aggregates over time the signal intensity gets to the point where it is strong enough to register as an event?

-Greg
UHN/Sickkids FCF
<<

Johanna

Participant

Posts: 8

Joined: Thu Jul 28, 2016 5:23 pm

Post Fri Oct 21, 2016 5:46 pm

Re: Strange signal shift

A short update:

In my tests yesterday I did not see anything comparable to what I observed last week. If I specifically look for it, for some of the markers the signal intensity for the negative population goes up very, very slightly in the course of an 45-minutes run, but very far from what I observed last week and not still clearly distinguishable from the positive population.

I should point out that I did not see high background or antibody streaking this time, and the cells I used were "healthier" than in my last experiment. Also, I did not use my whole antibody panel, but included only the "problematic" antibodies plus CD45, CD19 and CD3 to define B and T cells... But in general the problem seems to be that specific sample preparation/run rather than my general protocol.

I also wondered it could somehow be related to residual wash solution contaminant left in the SuperSampler syringe which was mixed into the sample upon initial aspiration and mixing and promoted sample degradation.

At the end of my run, with some sample left, I added a small amount of wash solution to the sample so see what this would do to the signals. As one would expect, background increased a lot and I saw streaking in some, but not all of the channels. Over time this lead to a decrease of CD43Nd150 signal on the positive population and a very mild signal increase for the negative population, resulting in reduced resolution of the population.
Unfortunately I could only run this for about 10 minutes since it was just leftovers of my sample and I was running out, so I cannot tell whether the signal would have further changed.
If my sample had been degraded over time due to residual wash solution, I think I would at least have initially also seen a decrease in staining of the positive population as observed here, which I did not. Also, I would expect to see this in the first run(s) only, since I did not wash between the single runs as I kept using the same sample.


I am sorry I could not shed more light onto this issue. If anyone has additional thoughts on this, I would be happy to here them. Since it does not seem to be a problem with my general sample preparation process, I will continue with my experiments next week. I will keep my eyes open, but hope this won't happen again.

Thank you all for your input!
Johanna
<<

BjornZ

Contributor

Posts: 43

Joined: Fri Jul 10, 2015 1:04 am

Post Fri Oct 21, 2016 6:39 pm

Re: Strange signal shift

Glad to hear that it was perhaps a transient issue.

Re: background subtraction, my apologies -- I mixed up Diva with the CyTOF software. I'm not sure now if there is a background subtraction similar to Diva's -- there's no option in the UI, and the manual doesn't mention any sort of running background subtraction. (I'm not that familiar with how "noise reduction" works and if that is comparable.) Furthermore you said the rain did not have high background, so, ignore my comment, sorry.

While it sounds like the wash solution hypothesis is not a very fitting explanation at this point (especially because the degradation should not have been cyclical), if you have further concerns about it, you can click the "flush" button before loading your sample, with the sample inlet tubing in the waste receptacle. That will eject 500 ul of water out of the inlet tube. This should not be necessary if you load a water tube after the cleaning cycle and preview long enough for the rain to be clear.

Best,
Zach
<<

GregBehbehani

Master

Posts: 85

Joined: Tue Apr 12, 2016 10:17 pm

Location: The Ohio State University, Columbus, Ohio

Post Fri Oct 21, 2016 7:01 pm

Re: Strange signal shift

Hi Johanna,

Thanks for the update. I agree, if you didn't wash between sample injections then the wash solution should not be the issue.

Based on your new results, I think background (and how the noise reduction algorithm compensated for it) might be the cause of what you had originally observed. I would definitely try re-extracting the IMD files from your original experiment with the "noise reduction" setting turned off (this box is checked by default, but you can un-check it). There are instruction for how to do this in the CyTOF manual, it should be in the section "IMD processing." If you compare the new FCS files that are created to the original ones, you may see a change in the signal shift, this might shed some light on what was going on.

Either way, I'm glad you're not having the problem any more.

best,

Greg
<<

vtosevski

Contributor

Posts: 44

Joined: Wed Nov 20, 2013 12:50 pm

Location: Zurich, Switzerland

Post Mon Nov 07, 2016 4:19 pm

Re: Strange signal shift

Dear all,

I wanted to reinvigorate the discussion and share our observerations on the same/similar issue.

We've been seening the similar phenomenon appearing from time to time - different samples (mouse, human), different labs, different panels, different buffers, different instruments. The problem has manifested itself as both the drop in the staining intensity of the positive populations and as an increase in the intensity of the negative populations.

We've tried many different angles at this but nothing we came up with so far could account for it - we've been seeing it with SuperSampler and PSI, with or without washing in between, with different data processing approaches etc.

We tried looking into water quality as well, comparing our home-made water and Fluidigm stocks - we found issues in both. I am attaching some plots below...

If anyone has any updates on this, or new ideas, please bring them forward - the problem still persists.

Vinko
Attachments
161010 new water data comparisons.pdf
(669.49 KiB) Downloaded 449 times
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Mon Dec 05, 2016 5:20 pm

Re: Strange signal shift

Hi all,

I finally followed up with Fluidigm regarding the Noise Reduction that's checked "on" by default in the software.

The response I got:
"The algorithm calculates the signal in between of events (average per push per channel) and if enabled - subtracts the value (multiplied by the next event duration) from the next found event signal."


So, in a well-behaved sample, this would have minimal effect on your data: you would have very litle or zero signal between cell events, and so would be subtracting zero/very small from your Cell events. In a sample with a streak (debris, streaky antibody, etc), though, this could be expected to affect only some channels (those with the higher background).

This doesn't address Johanna's statement that her affected channels didn't have noticeable streaking, though.....this was only to address Zach and Greg's (and my) discussion about what Noise Reduction actually does.


Mike
<<

kunicki

Contributor

Posts: 38

Joined: Thu Apr 13, 2017 8:46 pm

Post Fri Mar 30, 2018 6:35 pm

Re: Strange signal shift

Hello,

I'm seeing a similar issue, and found this discussion very helpful with troubleshooting it! Thank you everyone for sharing your input here.


The dramatic shift seems limited to the amu 141-149 range on our machine, where we typically see higher background signal after a HF wash. Like the other two operators though, I've continued to see the shift in signal intensity even when no HF wash was used prior to running the sample, and it's a transient between experiments. This would mean I've run cells/water through the lines of the SuperSampler for at least 30 minutes, after a short HF wash, and still see the effect when a new sample is run.

I would imagine replacing specific parts in the instrument could help ID the largest source of the specific background, but any other thoughts as to why this issue is transient, would be much appreciated!

Matthew
<<

mccars

Participant

Posts: 4

Joined: Wed Jun 15, 2016 10:56 am

Post Mon Jul 09, 2018 9:47 am

Re: Strange signal shift

Hello, has anyone found a solution for this strange signal shift?
I am currently having a similar problem. In my sample staining it appears as if the negative signal is becoming positive and vice-versa (see attached figures).

I work with PBMCs, and have included a Technical Healthy Control (PBMCs isolated from a blood cone from one individual) in every CyTOF run. I have recently returned from extended leave and wanted to ensure little or no batch effect in a large cohort of samples I’ve been processing.

I ran a QC bridging experiment on the technical controls and initially detected the problem in my T-cell panel only and thought the problem may be related to PMA/Ionomycin stimulation, but now I’m seeing it in my PBMC panel as well.

My protocol for my T-cell panel is quite like Johanna’s. Cryopreserved PBMCs are allowed to rest for 30 mins prior to 4 hr stimulation with PMA + Ionomycin plus BrefeldinA/Monensin. 103Rh viability staining for 15 mins at end of stimulation then cell surface staining, followed by ebiosciences FOXP3 permeabilisation/fixation protocol, and then intracellular staining. The samples are barcoded and then fixed again using MaxPar Fix/Perm, plus 100 uL of 1.6% PFA and left overnight before running on the Helios.

For the PBMC panel samples are allowed to rest for 30 mins before cell surface staining and barcoding, then MaxPar fixation as above.

The effect is seen in the following channels

T-cell panel:
142Nd, 144Nd, 145Nd, 146Nd
150Nd, 154Sm, 155Gd, 158Gd
173Yb, 175Lu

PBMC panel:
141Pr, 145Nd
153Eu

Possibilities that have been ruled out:
Other colleagues do not get this signal shift with the same markers, so it is not the machine.
I have also run the same experiment with a new technical control in parallel, and the results were the same, so it’s not sample specific.

At the moment our working theory is that this may be something to do with inadequate fixation, however my MaxPar Fix/Perm 3 months old and we add 1.6% PFA.

Please let me know if you have any ideas,
Thanks! Sheila.
Attachments
Signal.shift.bridging.expt.pdf
(555.48 KiB) Downloaded 405 times
PreviousNext

Return to CyTOF troubleshooting

Who is online

Users browsing this forum: No registered users and 7 guests

cron