histoCAT++ manual

Looking for ways to analyze your dataset? Need help with a software package?
Computational image analysis of multiplexed Imaging Mass Cytometry acquisitions
RaulCatena
Posts: 30
Joined: Mon Nov 27, 2017 12:49 pm

Re: histoCAT++ manual

Post by RaulCatena » Fri Apr 20, 2018 2:27 pm

One more thing,

If you download the last version that I just uploaded a minute ago, you can activate the compensation checkbox and the matrix published in Chevrier´s/Zanotelli´s paper is applied. You can further customize the compensation matrix by going to the edit->compensation matrix... menú. You can manually input new spillover values provided that you have calculated your own spillover matrix. Changes are saved in the workspace.

Cheers,

Raúl
RaulCatena
Posts: 30
Joined: Mon Nov 27, 2017 12:49 pm

Re: histoCAT++ manual

Post by RaulCatena » Fri Apr 20, 2018 3:14 pm

Found the problem regarding the TXT files after compensation.

The line returns in those files as they come from the CyTOF software are \r\n , which is the typical for windows. The file you generated had only \n (linux and mac), which led to a parser failure. Just patched the version (you can already download it) to handle that. However, when in doubt, make sure the scripts separate lines with \r\n so that we render files compatible to Fluidigm's native TXT files.

Update. Vito has updated the scripts so that the resulting TXT compensated file has the windows style \r\n line separator. In general, the R function fwrite defaults different depending on the OS. just passing the parameter eol = "\r\n" forces the right line separator to keep consistency in any OS.

One final comment, the current compensation in histoCAT++ is a fast approximation to that of Stephane's publication with Vito's scripts. In the future, the very same formulas will be implemented in histoCAT++ as well. In the meantime, you should find it convenient for quick exploration.

Raúl
ssivajothi
Posts: 33
Joined: Thu Jan 11, 2018 9:00 pm

Re: histoCAT++ manual

Post by ssivajothi » Mon Apr 23, 2018 2:21 pm

Thank you Raul and Vito for the quick resolution.

It was interesting to note that the spillover matrix I generated in-house is very similar to a 'purity matrix' file I received from Fluidigm a long time ago. However, the matrix from Stephane's paper has consistently higher spillover values. I would like to ask, how sensitive are those values to real world results in terms of images?
For example, between spillover counts of 0.2 and 0.5, how much real imaging differences can you see? (assuming an abundantly expressed marker). I guess this is to understand what the sensitivity /error margin is in the matrix construction. Interested to hear your experience.

Santhosh
RaulCatena
Posts: 30
Joined: Mon Nov 27, 2017 12:49 pm

Re: histoCAT++ manual

Post by RaulCatena » Mon Apr 23, 2018 10:45 pm

Hi Santhosh,

The spillover matrix could be quite different depending on the instrument, tuning, etc. However, by looking to our images and a few others from other labs and simply applying Stephane´s matrix (the so-called factory matrix in hCAT++) it does a very good job off-the-bat. Reason is that often spillover is close to noise, so an additional percentage point spillover is most of the times negligible. In general, I´d say that best is to apply your own matrix (Either with Stephane´s-Vito´s scripts or with histoCAT++), second best is to use Stephane´s matrix with the paper scripts, third is histoCAT++ factory settings (stephane´s matrix but slightly different calculations), worst is to do nothing. But again, from worst to best you find slight improvements in each step. At the end of the day, spillover in MC exists, but is not too massive.

Cheers,
MariekeIJsselsteijn
Posts: 10
Joined: Mon Jan 15, 2018 10:52 am

Re: histoCAT++ manual

Post by MariekeIJsselsteijn » Tue Apr 24, 2018 10:04 am

Hi all,

I noticed that there is an option to export the cell data as an FCS file but its 'greyed out'. is this option not available yet or is something else needed to be able to do this. alternatively, does anyone have an idea how to convert tsv/txt files into FCS files.

It would be great to be able to move the cell data to programs like cytosplore for in depth HSNE analysis next to the spatial data from the images.
RaulCatena
Posts: 30
Joined: Mon Nov 27, 2017 12:49 pm

Re: histoCAT++ manual

Post by RaulCatena » Tue Apr 24, 2018 7:45 pm

Hi Marieke,

The export FCS function will be available in the near future. Actually, tSNE will be included too, but not immediately.
Summarizing, to go down the tSNE road we have the following roadmap.

1) Currently: Export TSV from hCAT++, then use R or Python scripts to convert to FCS, and good to go to Cytoxplore, Accense, etc. (or do tSNE with R or Python scripts directly).
2) FCS function in histoCAT++ ready in the near future. Then Cytoxpore, etc...
3) Do everything directly in histoCAT++´s next version, but that may take a little more time, cannot determine right now how long but I´ll keep you up to date.

Regards,

Raúl
DenisSchapiro
Posts: 41
Joined: Wed Nov 29, 2017 11:34 am

Re: histoCAT++ manual

Post by DenisSchapiro » Tue Apr 24, 2018 8:02 pm

Dear Marieke,

Another option would be to use histoCAT which has a fcs/csv export function.

Best

Denis
RaulCatena
Posts: 30
Joined: Mon Nov 27, 2017 12:49 pm

Re: histoCAT++ manual

Post by RaulCatena » Thu Apr 26, 2018 10:06 am

There is an FCS export option for the cell data in histoCAT++ working now. Download the newest version if you want to use this.
The generated FCS files can be imported in FlowJo, Accense, flowCore and other tools. For some reason do not work on Cytobank, we will look into that at some point.
To generate the FCS file simply go to the 'Cell Data' option in the top left menu (on top of the left table), select the cell data object you want to export as FCS, select in the channel column the channels that you want to export (don't forget this, else you'll get something empty), and click export as FCS or TSV, as you wish.

Best,

Raúl
ssivajothi
Posts: 33
Joined: Thu Jan 11, 2018 9:00 pm

Re: histoCAT++ manual

Post by ssivajothi » Fri Jul 06, 2018 7:14 pm

Hi,

I just wanted to ask if there is going to be any update to the Histocat++ manual covering data analysis?

Also, are Histocat and Histocat++ going to be developed separately?

Thank you,
Santhosh
RaulCatena
Posts: 30
Joined: Mon Nov 27, 2017 12:49 pm

Re: histoCAT++ manual

Post by RaulCatena » Fri Jul 06, 2018 9:30 pm

Hi Santhosh

During the summer I'll update a manual on Analysis. It will be corresponding to the latest version released as of May. Regarding future developments of both histoCAT and histoCAT++ please contact Bernd Bodenmiller.

Regards,

Raúl
Post Reply