FAQ  •  Register  •  Login

Phenograph

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 :-)
<<

GregHopkins

Contributor

Posts: 26

Joined: Tue Apr 11, 2017 8:39 pm

Location: 2seventy bio, Cambridge, MA

Post Wed Jan 31, 2018 4:07 pm

Phenograph

Hello,

I am not famliar at all with Matlab or Cyt, but would like to start running some analysis of my data in Phenograph. Is anyone here from the Boston area that may be able to help me at some point (or someoene elsewhere able to do a conference call) and show me the basics on running Phenograph?

Regards

Greg Hopkins
<<

bc2zbUVA

Contributor

Posts: 22

Joined: Thu Nov 19, 2015 4:23 pm

Post Wed Jan 31, 2018 4:12 pm

Re: Phenograph

CyTOFkit includes phenograph if you can get R up and running.

https://www.bioconductor.org/packages/3 ... ample.html
<<

vtosevski

Contributor

Posts: 44

Joined: Wed Nov 20, 2013 12:50 pm

Location: Zurich, Switzerland

Post Thu Feb 01, 2018 8:59 am

Re: Phenograph

Hi Greg,

With the most basic R literacy you can run the standalone Rphenograph package available from here:

https://github.com/JinmiaoChenLab/Rphenograph

Vinko
<<

MCOlivier

Contributor

Posts: 47

Joined: Mon Oct 05, 2015 9:48 am

Location: European Genomic Institute for Diabetes, Universitity of Lille, Institut Pasteur de Lille, France

Post Thu Feb 01, 2018 11:04 am

Re: Phenograph

Hi,

I just want to share my experience with CytofKit, which proposes 3 different clustering algorythms: Rphenograph, ClusterX and DensVM.
After a quiet high number of runs on those 3 algo, I really found ClusterX as the most accurate to identify well delineated clusters after tSNE. This is "mathematically explainable", as amongst those 3 algo, only Cluster X bases it's clustering on the 2 tSNE avlues, while both Rphenograph and DensVM use the parameters entered upstream to viSNE dimension reduction
If you use the the GUI interface in CytofKit, it's really easy to run ^^

Hope this will help.

Olivier
<<

tomash

Contributor

Posts: 25

Joined: Sun Oct 19, 2014 10:15 pm

Post Thu Feb 01, 2018 11:16 pm

Re: Phenograph

Hi Greg,

We'd be happy to do a conference call to walk you through whichever option you prefer. We are in Sydney, Australia (US east coast + 16 hrs) so we'd just have to line up a time. Shoot us an email at thomas.ashhurst@sydney.edu.au if you want to set it up. However, before we do that, you should check out the resources available to help get you going.

PhenoGraph is an awesome tool, I'm a huge fan. Both the Cyt package in Matlab and the CytofKit package in R, described in the other comments, can run in a user-interface format (point and click, rather than actual coding) which is a little easier to use. There is still a bit of learning involved to get it to run properly, but it's not a huge threshold to get past. Then there is the scripted version that you can run in R using the package mentioned by Vinko. This is the 'better' way to do it, because you have a huge amount of flexibility in managing your data. Whilst it is a little harder to learn, it is a huge investment in data analysis in the future. There is also some other options out there, many of which are included in the CytofKit package. I've been particularly partial to FlowSOM these days for running really large datasets (https://github.com/SofieVG/FlowSOM) but we run it in R. CytofKit contains FlowSOM code, but I don't think that it is implemented in the user interface component at the moment (though I might be wrong, anyone reading feel free to correct me).

If you download the Cyt package from Dana Pe'er's website (https://www.c2b2.columbia.edu/danapeerl ... nload.html) there is a screenshot guide that comes with it. We also have a screenshot protocol at https://sydneycytometry.org.au/computational-analysis/. Our current version is focused on running tSNE in Cyt, but many of the steps are in common. Because of the way we store it, you'll have to 'request permission' after clicking on the link. One thing you'll have to keep an eye out for in Cyt is a that in your FCS files, some parameters will come with a 'parameter' name but not a 'stain' name (such as event length, etc). These come up as blank stain names in Cyt which stop the program working properly. You can use the re-naming function in Zach Bjornson's R package (https://github.com/nolanlab/cytofCore), or just avoid including parameters that would have the problem (event length, time, or FSC/SSC etc from flow data). We have some screenshots and an implementation of Zach's script if you want to try it -- it's not online yet, but I'll throw it up on https://github.com/sydneycytometry later today. The CytofKit guide (https://www.bioconductor.org/packages/3 ... ample.html) has a bunch of screenshots to walk you through using it, so that should be fairly straight forward.

Try these out first, and see how far you get. Feel free to get in touch in the meantime if you need help.
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Thu Feb 01, 2018 11:29 pm

Re: Phenograph

Hi Tom,

In my work with CytofKit (for the Multicenter paper, among others), you are correct: FlowSOM is *not* something you can select in the initial GUI setup (only ClusterX, DensVM, and Phenograph), at least not up through v1.4.10.

However, once you get your results, FlowSOM is *then* something you can access from the results GUI: see Section 3/3 here: https://www.bioconductor.org/packages/r ... nyAPP.html


Maybe one of the Cytofkit maintainers will move it to the front GUI in a later version? ;)


Mike
<<

ErinSimonds

Master

Posts: 50

Joined: Tue May 13, 2014 8:04 pm

Post Thu Feb 01, 2018 11:43 pm

Re: Phenograph

Hi all,

FlowSOM is available as an option in the GUI in the current version of Cytofkit on Github (v1.11.3), but not the version on Bioconductor (v1.10.0). Screenshot of the current GUI attached. This also runs tSNE using the Rtsne package, so there's no need to use Matlab / cyt for this step.

To install it from Github, paste these lines into R:

  Code:
if(!require(devtools)){
  install.packages("devtools") # If not already installed
}
devtools::install_github("JinmiaoChenLab/cytofkit")


More info here:
https://github.com/JinmiaoChenLab/cytofkit

Help requests and bug reports can be filed on Github in the "Issues" section:
https://github.com/JinmiaoChenLab/cytofkit/issues

ES
Attachments
180125 FlowSOM1 setup.png
<<

dtelad11

Master

Posts: 129

Joined: Mon Oct 31, 2016 6:26 pm

Post Thu Feb 01, 2018 11:48 pm

Re: Phenograph

I will second Erin's suggestion, go with FlowSOM. It is faster by an order of magnitude and leads to results which are comparable to Phenograph.
<<

tomash

Contributor

Posts: 25

Joined: Sun Oct 19, 2014 10:15 pm

Post Fri Feb 02, 2018 12:01 am

Re: Phenograph

Ah, thanks for the clarification Erin -- evidently I have the Bioconductor version at the moment. We've been using FlowSOM extensively in our own code so I haven't been following the CytofKit updates. I'm glad it's in there, makes such a different for large datasets. Thanks Mike as well for the tip about running it after the results come back!

Tom
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Wed Mar 21, 2018 5:14 pm

Re: Phenograph

Hi Erin,

I've been trying to update my Cytofkit to the v1.11.3 you mentioned in early Feb. I'm installing on a Mac Pro, recently updated to High Sierra (10.13.3), R updated to 3.4.4, and RStudio updated to 1.1.442. XQuartz was updated after the update to High Sierra, and is now X11 2.7.11.

When I tried your code to install directly from Github, I first tried it in RStudio, and got prompted to install XCode (the link out is to a CNet page: https://www.cnet.com/how-to/install-com ... s-in-os-x/).
Installation fail in RStudio1pt1pt442.png


(you can also see in the screenshot that both cytofkit 1.10.0 and devtool packages are installed)

I eventually got XCode Version 9.2 installed, and tried it again. Same failure, and loop out to the CNet page.

So, I tried it directly in R, and it also failed, but with different error message:
Installation fail in R3pt3pt4.png



Do you have an idea of what might be happening here?

I tried looking on the Github Cytofkit site for a tar.gz to install it directly, but couldn't find that either.


Mike
Next

Return to CyTOF data analysis

Who is online

Users browsing this forum: No registered users and 13 guests