FAQ  •  Register  •  Login

Data normalization with EQ beads

Forum rules
Please be as geeky as possible. Reference, reference, reference.
Also, please note that this is a mixed bag of math-gurus and mathematically challenged, so choose your words wisely :-)
<<

Ofir

Master

Posts: 75

Joined: Thu Nov 07, 2013 12:46 pm

Location: US, CA

Post Thu Mar 13, 2014 9:16 pm

Data normalization with EQ beads

Hey gang,
What software do you use to normalize CyTOF data using EQ Four Element Beads?
If there's more than one option, I would appreciate pros and cons.
<<

mleipold

Guru

Posts: 5792

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Thu Mar 13, 2014 11:59 pm

Re: Data normalization with EQ beads

Hi Ofir,

I'm only aware of the Nolan lab's normalization algorithm (Matlab-based). That's what the HIMC uses, when we use norm beads.

Note: while the default settings for the program *are* the custom beads from their paper, you can select the current 4-element bead isotopes instead.


Mike
<<

ianfrank

Contributor

Posts: 29

Joined: Thu Nov 21, 2013 11:59 pm

Location: Tampa, FL

Post Fri Mar 14, 2014 6:14 pm

Re: Data normalization with EQ beads

Ofir,

The new Cytof 2 software has the normalization function in it. I think that DVS is probably willing to share the software. We have the first generation Cytof, but I've installed the Cytof 2 software on a separate PC. I was warned not to install the Cytof 2 software on the PC attached to the original Cytof. I've only done a trial run, but it seemed pretty user friendly and easy to install.

Ian
<<

robs

Contributor

Posts: 40

Joined: Mon Dec 02, 2013 3:42 pm

Location: University of Connecticut

Post Mon Mar 17, 2014 3:20 pm

Re: Data normalization with EQ beads

I don't think they would want me to upload the install file here, but if you make a my DVS account http://www.dvssciences.com/mydvs/ you can get a copy of the newest software 6.0.626 with the easy to use normalizer. Its under CyTOF2 > software. The instructions are in the software release notes on the same page.
To make a My DVS account, you just need your CyTOF's serial number.
The nice thing about their normalizer is that it is lot specific and normalizes back to a factory standard.
Rob
<<

AdeebR

Grand master

Posts: 169

Joined: Thu Mar 13, 2014 5:58 pm

Location: NYC

Post Wed Mar 19, 2014 1:36 am

Re: Data normalization with EQ beads

I've used both and find the CyTOF one slightly more convenient because it's more automated; the Nolan lab tool gives the user more control over the process. One thing that I prefer about the Nolan lab tool is that the normalized data files are ultimately saved to a new folder, which is a little more organized than the CyTOF tool which just appends the file name with a "_1" and leaves them in the same folder (this might be a modifiable setting).

I know that bead-based normalization should theoretically allow us to normalize all CyTOF data across instruments, which would make data sets from different studies and institutions directly comparable and very amenable to meta analyses (assuming the cell preparation and staining were also standardized). Very cool if it works. Do others think the Eq beads will actually allow this to happen or are there other issues that would need to be addressed?

One clarification regarding the Nolan lab tool: since it doesn't use lot-specific bead information, does it ultimately only work to normalize the files across a single run (i.e., files that you input as part of a single normalization), or is the normalized data generated with this tool also valid over time and across instruments?
Adeeb Rahman
Icahn School of Medicine at Mount Sinai, NYC
<<

mleipold

Guru

Posts: 5792

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Wed Mar 19, 2014 2:58 pm

Re: Data normalization with EQ beads

My experience is that prep-to-prep variation among in-house conjugates can blur these lines a bit. Sometimes you have a good prep, sometimes you have a great prep, sometimes you have a borderline prep.

It all depends on what your ultimate goal is: in our case, we're mainly just wanting to resolve our populations of interest, so as long as we can do that sufficiently well, we don't care quite so much about absolute staining intensities (ie, that the median Dual counts for Population X are exactly the same from one batch of antibody to the next). This poses an issue for archival normalization (including your meta analyses): if between one project and the next (or even worse, one plate to the next), you switch a "lot" of only one antibody of your cocktail and its performance changes, then normalization will NOT "fix" that.

So, since many people currently still use a number of in-house conjugates in their panels, this will continue to be a small limitation on normalization.


Re Nolan tool: it should allow you to normalize across all files that use that particular lot of beads.
<<

DAVID

Participant

Posts: 17

Joined: Tue Nov 26, 2013 1:55 pm

Post Wed Aug 13, 2014 8:29 am

Re: Data normalization with EQ beads

Back to the original question...has anyone compared the two normalisation approaches head to head? i.e. are the algorithms for the MATLAB and DVS normalisers based on the same principle(s)?

Is there a way to get graphical output from the DVS CyTOF2 software after normalization to see how the signal intensities have been changed for each file?

Just a note which could help with your data preprocessing workflow...after we removed some channels using other software (e.g 'R' or 'Infinicyt') the CyTOF2 normalizer didn't work anymore. The Matlab software works fine with our channel-edited files, and although this software (Matlab) produces a graphical output at the end...it is difficult to customise the image settings to see all the details/filenames on the graph when you have normalised large numbers of files in one batch (150 to 300, but even for smaller experiments of 50 samples).

Also the DVS software automatically downloads a Bead Lot number from the internet which is slightly different to the label on our Bead bottle...the bead bottle has an extra number 4 after the letters EQ... I'm guessing (and hoping) that it's the same lot but will confirm with DVS.

Any info / advice would be much appreciated, especially about the first point.

Cheers,

David
<<

anitamkant

Master

Posts: 51

Joined: Mon Nov 18, 2013 6:30 am

Post Wed Aug 13, 2014 8:17 pm

Re: Data normalization with EQ beads

1) Please Read the user guide that will give you a head on comparison and principle behind the beads and SW. Matlab software and CyTOF SW work differently.
2) Graphical output: Fluidigm has noted the feedback and will take action appropriately.
3) @ David: Can you explain which channels were taken off in your workflow?
4) Contact Tech support at DVS to upgrade for the SW.
5) Please register at http://www.dvssciences.com to get an access to the portal for important information; User Guides; Panel Designer Tool.
<<

DAVID

Participant

Posts: 17

Joined: Tue Nov 26, 2013 1:55 pm

Post Mon Aug 18, 2014 11:57 am

Re: Data normalization with EQ beads

Thanks Anita,

For point 3: We removed channels 131, 138, 190. Though I think the problem might be caused by trying to normalise data from both CyTOF1 and CyTOF2 in the same batch using the DVS CyTOF2 software, has this worked for anyone else? We get a 'bead normalization error (-4)' when the software tries to normalise cytof2-generated data.

David
<<

anitamkant

Master

Posts: 51

Joined: Mon Nov 18, 2013 6:30 am

Post Tue Aug 19, 2014 4:27 am

Re: Data normalization with EQ beads

Hi David, Can you approach tech support with your data files? We would love to investigate further. Thanks

Return to CyTOF data analysis

Who is online

Users browsing this forum: No registered users and 7 guests

cron