flowAI: flow rare check in R vs FlowJo
1
0
Entering edit mode
Olena • 0
@a1384ece
Last seen 10 months ago
Germany

Dear All,

I just started to work with flowAI plugin and wondering if someone could explain the difference in analysis which I got by running flowAI in FlowJo and in Rstudio. I am using simply flow_auto_qc function to run analysis for all fcs files available in directory, which represent one batch. So, I am not specifying any parameters neither in R nor in FJ and using them by default. However, seems the flow rate check is more sensitive for outliers in FJ. It detects almost doubled amount of events as anomalies compared to flow rate check performed in R, - ignores many of those outliers defined during FJ session.

I am attaching these two pictures and would be glad to hear any idea/hints/suggestions. Because I am really confused, which way should be preferable for analysis.

Best, Olena

UPD: I performed a few more tests. And really cannot get how the difference in flow rate can be so huge. For some samples, flow_auto_qc removes up to 44% of anomalies (most of them detected during flow rate check), meanwhile flow rate for same sample in FlowJo has 0% of anomalies. I could identify only one parameter which is different for FJ and R package, it is pen_valueFS, - 200 by default in FJ and 500 in R package. However, the higher this value is, the less strict detection of anomalies should happen, what doesn't seem to be valid in the present situation. Setting this value to 500 in FJ doesn't make the outcome similar to R package.

As previously, it would be super helpful if someone has any idea of what happens here. here is a screen from them

flowAI • 907 views
ADD COMMENT
0
Entering edit mode

Hi Olena! Have you figured out the reason behind this difference between FlowJo and R plugins?

I am experiencing the same issue :)

ADD REPLY
0
Entering edit mode
mongianni1 • 0
@mongianni1-23386
Last seen 3 months ago
Boston

Hi both,

the problem is that flowJo has implemented an older version of flowAI and they did not update it for a while.

I recently updated flowAI to better handle the slow increase or dicrease of the flow rate. You should mainly use "alphaFR" to tune the amount of outliers to detect, and I also added the parameter "ModeDevFR" which allows to remove all the events at a certain deviation from the mode before performing outlier detection.

You should try the flowAI version from the devel branch (https://www.bioconductor.org/packages/devel/bioc/html/flowAI.html).

Eventually, I might ask the developers from flowJo to update the flowAI version that they use.

Best, Gianni

ADD COMMENT

Login before adding your answer.

Traffic: 428 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6