FAQ  •  Register  •  Login

Strange FCS files resulting from compensation with CATALYST

<<

PaulineM

Participant

Posts: 14

Joined: Thu Apr 12, 2018 1:05 pm

Post Fri Sep 14, 2018 2:19 pm

Strange FCS files resulting from compensation with CATALYST

Hello,

I compensated cyTOF data with CATALYST.
For most of my samples it worked perfectly.
Nevertheless, for some FCS files, I obtained strange resulting FCS files, with many events at zero (as expected), but also many negative events and no more cell population could be visualized (see picture of a successfully compensated file (left) and a compensated file with problem (right)).

I wonder if someone has already experienced such a problem and has found a solution to solve it.

Pauline
Attachments
screenshot.png
<<

mleipold

Guru

Posts: 5792

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Fri Sep 14, 2018 3:14 pm

Re: Strange FCS files resulting from compensation with CATAL

Hi Pauline,

Is there anything different about those two files (successful vs unsuccessful)?

For example
1. Did the successful one have a lot more cells than the unsuccessful one?
2. Is one acquired as one sample, while the other is concatenated or something?
3. What are the conditions of the experiment for the two files? For example, is one stim and the other unstim?
4. Are you looking at total cells, or have these been gated down somehow?
5. Are these the same sample type (eg, both PBMCs), or different (eg, one is PBMCs, the other is purified T cells or something)?

For the unsuccessful one: FlowJo can add additional negative decades, and additional positive decades. How many positive and/or negative decades do you have to add in order to push the events completely visible (ie, completely back off the axis)? That might give us info on what magnitude of problem you have.


Mike
<<

BjornZ

Contributor

Posts: 43

Joined: Fri Jul 10, 2015 1:04 am

Post Fri Sep 14, 2018 6:13 pm

Re: Strange FCS files resulting from compensation with CATAL

Hi Pauline,

Without knowing anything about CATALYST, I can tell you with almost-certainty that that's an improperly encoded FCS file (the floating point values are denormals, likely because the data coordinates are off by a byte or two, which might be from an unescaped delimiter in the header). I'd open an issue in their GitHub repository https://github.com/HelenaLC/CATALYST. You can tag me with @zbjornson there and I can chime in if needed.

-Zach
<<

sgranjeaud

Master

Posts: 123

Joined: Wed Dec 21, 2016 9:22 pm

Location: Marseille, France

Post Fri Sep 14, 2018 7:57 pm

Re: Strange FCS files resulting from compensation with CATAL

Hi,

Could you post the version of the flowCore package?

  Code:
packageVersion("flowCore")


Moreover you can try the following commands replacing "the_files.fcs" by the path to your FCS file, and post the result:

  Code:
con <- file("the_file.fcs", open="rb")
dat <- readBin(con = con, what = "raw", n = 8192, size = 1)
close(con)
rawToChar(dat)



Best.
<<

PaulineM

Participant

Posts: 14

Joined: Thu Apr 12, 2018 1:05 pm

Post Mon Sep 17, 2018 3:18 pm

Re: Strange FCS files resulting from compensation with CATAL

Hi Mike, Zach and Samuel,

Thank you for your answers;

Mike,

I run many tests. At first I noticed that the normalization had an impact on the output in catalyst. But then I run other tests with FCS files which have had exactly the same ‘treatment’ and still the problem persists for some FSC files.

For the last test that I ran:
1.  The number of cells did not affect the output
2.  I did not have any concatenated files among my files.
3.  The conditions were exactly the same for the different files.
4.  I did not do any ‘pre-gating’ on these files
5.  They were the same sample type (both PBMCs)

On FlowJo I can see that they are events below the 3rd negative decade and above the 13th positive decade (See picture)

Since my first post: I tried to solve the problem by re-writing the FCS files with updatePanel (on R) before running the compensation with CATALYST. It worked for some files, but not all of them.
So I tried re-writing the FCS files with updatePanel (on R) AFTER running the compensation with CATALYST. It worked for all the files I tested so far.
IT SEEMS THE PROBLEM IS SOLVED (I’ll let you know if I encounter another problem)

I am still very curious to know if other users encountered this issue or if I am the only one.

Zach,

Even though, I might have a solution to solve my problem now, I opened an issue in their GitHub repository as you suggested and tagged you

Samuel,

Here are the ouputs for the 2 commands:

> packageVersion("flowCore")
[1] ‘1.46.2’

> con <- file("/Users/maby/Desktop/compensation-test/test compens/original files/withViabLow_1b1.fcs", open="rb")
> dat <- readBin(con = con, what = "raw", n = 8192, size = 1)
> close(con)
> rawToChar(dat)
Error in rawToChar(dat) :
caractère nul au milieu de la chaîne : 'FCS3.0 58 7484 748582727492 0 0|$FIL|withViabLow_1b1.fcs|$TOT|328254|$PAR|63|$BYTEORD|4,3,2,1|$DATATYPE|F|SampleID|1b1|FJ_FCS_VERSION|3|$BEGINANALYSIS|0|$ENDANALYSIS|0|$BEGINSTEXT|0|$ENDSTEXT|0|$NEXTDATA|0|FCSversion|3|CREATOR|PengQiu FCS writer|$COM|PengQiu FCS writer|FILENAME|20171026_IB1_02_1_normalized.fcs|GUID|1.fcs|ORIGINALGUID|1.fcs|$MODE|L|$BTIM|15:58:32|$ETIM|16:20:09|$CYT|DVSSCIENCES-CYTOF-1.0.0|$DATE|26-Oct-2017|$CYTSN|cytof2|$P1S|190BCKG|$P2S|137Ba|$P3S|138Ba|$P4S|209Bi_CD16|$P5S|140Ce_EQBeads1|$P6S|142Ce_CD19|$P7S|133Cs|$P8S|161Dy_CTLA4|$P9S|162Dy_FOXP3|$P10S|163Dy_BTLA|$P11S|164Dy_CD95|$P12S|166Er_CD163|$P13S|167Er_CD27|$P14S|168Er_CD206|$P15S|170Er_CD3|$P16S|151Eu_ICOS|$P17S|153Eu_TIM3|$P19S|155Gd_PD1|$P20S|156Gd_CD49a|$P21S|158Gd_OX40|$P22S|160Gd_CD14|$P23S|165Ho_CD45RO|$P24S|127I|$P25S|115In|$P26S|191Ir_ADN1|$P27S|193Ir_ADN2|$P28S|139La|$P29S|175Lu_PDL1|$P30S|176Lu_CD127|$P31S|143Nd_CD5|$
>
Attachments
Capture d’écran 2018-09-17 à 16.13.45.png
<<

sgranjeaud

Master

Posts: 123

Joined: Wed Dec 21, 2016 9:22 pm

Location: Marseille, France

Post Mon Sep 17, 2018 7:13 pm

Re: Strange FCS files resulting from compensation with CATAL

Dear Pauline,

Thanks for your reply. Unfortunately, the output seems to be cut.
Maybe you should try to check the FCS files with flowIO (http://bioinformin.cesnet.cz/flowIO).

Best.
<<

PaulineM

Participant

Posts: 14

Joined: Thu Apr 12, 2018 1:05 pm

Post Tue Sep 18, 2018 5:20 pm

Re: Strange FCS files resulting from compensation with CATAL

Thank you very much for your help.

Here is the proper R command:
> con <- file("original files/withViabLow_9b1.fcs", open="rb")
> dat <- readBin(con = con, what = "raw", n = 8192, size = 1)
> close(con)
> rawToChar(dat)
Error in rawToChar(dat) :
caractère nul au milieu de la chaîne : 'FCS3.0 58 7388 738933835112 0 0|$FIL|withViabLow_9b1.fcs|$TOT|134237|$PAR|63|$BYTEORD|1,2,3,4|$DATATYPE|F|SampleID|9b1|FJ_FCS_VERSION|3|$CYT|DVSSCIENCES-CYTOF-1.0.0|$CYTSN|cytof2|$DATE|12-Dec-2017|$BTIM|17:45:42|$ETIM|18:02:44|$MODE|L|$BEGINSTEXT|0|$ENDSTEXT|0|$NEXTDATA|0|$BEGINANALYSIS|0|$ENDANALYSIS|0|$P1S|190BCKG|$P2S|137Ba|$P3S|138Ba|$P4S|209Bi_CD16|$P5S|140Ce_EQBeads1|$P6S|142Ce_CD19|$P7S|133Cs|$P8S|161Dy_CTLA4|$P9S|162Dy_FOXP3|$P10S|163Dy_BTLA|$P11S|164Dy_CD95|$P12S|166Er_CD163|$P13S|167Er_CD27|$P14S|168Er_CD206|$P15S|170Er_CD3|$P16S|151Eu_ICOS|$P17S|153Eu_TIM3|$P19S|155Gd_PD1|$P20S|156Gd_CD49a|$P21S|158Gd_OX40|$P22S|160Gd_CD14|$P23S|165Ho_CD45RO|$P24S|127I|$P25S|115In|$P26S|191Ir_ADN1|$P27S|193Ir_ADN2|$P28S|139La|$P29S|175Lu_PDL1|$P30S|176Lu_CD127|$P31S|143Nd_CD5|$P32S|144Nd_CD38|$P33S|145Nd_CD4|$P34S|146Nd_CD8a|$P35S|148Nd_GITR|$P36S|150Nd_LAG3|$P37S|189Os|$P38S|208Pb|$P39S|102Pd|$P40S|104Pd|$P41S
<<

PaulineM

Participant

Posts: 14

Joined: Thu Apr 12, 2018 1:05 pm

Post Wed Sep 19, 2018 6:13 pm

Re: Strange FCS files resulting from compensation with CATAL

Dear all,

PROBLEM IS SOLVED

If you ever get a similar problem as I had, I wanted to let you know that when I normalized FCS files with Matlab before running Catalyst it made catalyst work perfectly.

Pauline

Return to CyTOF general discussion

Who is online

Users browsing this forum: AdsBot [Google], Google [Bot] and 13 guests