combineExpression

[annecar]

Hi Nick,
I am running into issues while combining my scRepertoire data with my Seurat object. The description of the function includes "By default this function also calculates the frequencies and proportion of the clones by sequencing run". I a am not sure this makes sense for my purpose. How can I have the global frequency of the clonotypes over the entire Seurat object ? Is that at all possible? I am not sure whether it makes sense to plot frequencies of batches on the UMAP for example, there I would like show frequencies of the whole repertoire. Is that possible?

Another question is whether it is possible to append the frequencies for IGL and IGH in one go. As I understand, I would need to do this separately. One could somehow go via Metadata, of course.

Thank you for your help and all the best!
Annecar

[ncborcherding]

Hey @annecar ,

How can I have the global frequency of the clonotypes over the entire Seurat object?

Probably the easiest way would be to make a universal variable across all the combined data and use that as a group by?

combined.BCR <- combineBCR(contigs, ....)
combined.BCR <- AddVariable(combined.BCR,  variable.name = "Universal", variables = rep("ID1", length(combined.BCR)))

I am not sure what your setup is, but I think it is important to think of your numerator and denominator for these calculations. The reason that the sequencing run is the default is that sequencing runs can vary vastly in the number of recovered clones; using that as the denominator is a decent initial assumption. Suppose you want to group several runs across the same individual or mouse sample. In that case, that might also be worth trying in lieu of the whole repertoire across multiple individuals/sequencing runs.

Another question is whether it is possible to append the frequencies for IGL and IGH in one go. As I understand, I would need to do this separately. One could somehow go via Metadata, of course.

You can play around with addLabel in combineExpression() see the reference. You can add the IGL and IGH in succession with new column headers and the respective frequencies.

Hope that helps,

Nick

[annecar]

Dear Nick, Thank you so much for the ideas. I will try those out. All the bets for now, Anne Anne Eugster, PhD Scientist, Research Group Binder Translational Immunology-Oncology Department of Biomedicine University Hospital Basel Hebelstrasse 20 4031 Basel Switzerland ***@***.*** On 4 Nov 2025, at 08:03, Anne Eugster ***@***.***> wrote: Anne Eugster, PhD Scientist AG Bonifacio Technische Universität Dresden Center for Molecular and Cellular Bioengineering (CMCB) Center for Regenerative Therapies (CRTD) Fetscherstr. 105 01307 Dresden Tel: +49 (0)351-458-821 41 Fax: +49 (0)351-458-821 09 Email: ***@***.******@***.***> Webpage: http://www.tu-dresden.de/cmcb/crtd/ From: theHumanBorch ***@***.******@***.***>> Reply to: BorchLab/scRepertoire ***@***.******@***.***>> Date: Friday, 31 October 2025 at 15:36 To: BorchLab/scRepertoire ***@***.******@***.***>> Cc: Anne Eugster ***@***.******@***.***>>, Mention ***@***.******@***.***>> Subject: Re: [BorchLab/scRepertoire] combineExpression (Issue #550) [Image removed by sender.]ncborcherding left a comment (BorchLab/scRepertoire#550)<#550 (comment)> Hey @annecar<https://github.com/annecar> , How can I have the global frequency of the clonotypes over the entire Seurat object? Probably the easiest way would be to make a universal variable across all the combined data and use that as a group by? combined.BCR <- combineBCR(contigs, ....) combined.BCR <- AddVariable(combined.BCR, variable.name = "Universal", variables = rep("ID1", length(combined.BCR))) I am not sure what your setup is, but I think it is important to think of your numerator and denominator for these calculations. The reason that the sequencing run is the default is that sequencing runs can vary vastly in the number of recovered clones; using that as the denominator is a decent initial assumption. Suppose you want to group several runs across the same individual or mouse sample. In that case, that might also be worth trying in lieu of the whole repertoire across multiple individuals/sequencing runs. Another question is whether it is possible to append the frequencies for IGL and IGH in one go. As I understand, I would need to do this separately. One could somehow go via Metadata, of course. You can play around with addLabel in combineExpression() see the reference<https://www.borch.dev/uploads/screpertoire/reference/combineexpression>. You can add the IGL and IGH in succession with new column headers and the respective frequencies. Hope that helps, Nick — Reply to this email directly, view it on GitHub<#550 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AIM7QPRB6FGGA4WRHNIOI2L32NXXRAVCNFSM6AAAAACKX7MPQ6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTINZTGM3TANJRGI>. You are receiving this because you were mentioned.Message ID: ***@***.******@***.***>>

[annecar]

Hi Nick again,

I knew I would run into some trouble again... I am still struggling with the combineExpression function. If I run it with chain="both", I obtain clonotypes like this:

IGHV1-69.IGHD3-10.IGHJ6.IGHM_IGLV3-25.IGLJ1.IGLC1
NA
IGHV1-46.NA.IGHJ3.IGHM_IGKV1-9.IGKJ3.IGKC
IGHV3-23.NA.IGHJ4.IGHG1_IGLV2-8.IGLJ3.IGLC2
NA
IGHV4-59.IGHD2-21.IGHJ3.IGHM_IGKV1-9.IGKJ1.IGKC

whereas if I run it with If I run it with chain="IGL", I obtain clonotypes like this:

IGHV1-69.IGHD3-10.IGHJ6.IGHM_IGLV3-25.IGLJ1.IGLC1
NA
NA
IGHV3-23.NA.IGHJ4.IGHG1_IGLV2-8.IGLJ3.IGLC2
NA
NA

It looks like the kappa chain clonotypes are not considered when using teh argument chain="IGL". Why is this, and how can I obtain frequencies for both types of light chains?
You suggested to subset to separate the light chains, but I can only do this once I added the scRepertoire data to the Seurat object. Right?

So that you understand better what I am after: I would like to have, in the end, for each cell, its IGH/IGL/IGK Clonotype info and the accompanying frequencies. Once I have this, I would like to make subsets of cells that have more than one light chain, and again obtain Clonotype and frequency.

So from this also the question derives how to recalculate the frequencies after subsetting.

I am sorry to ask so many things. But that you for your help!
By for now,
Anne

[ncborcherding]

Hey Anne,

Thanks for continuing to follow up - I think you found a small bug:

https://github.com/ncborcherding/scRepertoire/blob/040be3e1580700bd9d147f96c442c74c8bf545ea/R/combineExpression.R#L79-L81

Although combineExpression() is suppose to allow the frequency/proportion calculation by individual chains, the internal portion of the chain dissection called .offTheChain() has an internal filtering that was not not disabled here.

Just pushed the relevent fix to the main repo - working on a larger bug in Seurat at the moment, but this should work for your purposes.

Nick

[annecar]

Ok, thanks a lot!! Do I need to update the scRepertoire package? Anne Eugster, PhD Scientist, Research Group Binder Translational Immunology-Oncology Department of Biomedicine University Hospital Basel Hebelstrasse 20 4031 Basel Switzerland ***@***.*** On 6 Nov 2025, at 16:59, theHumanBorch ***@***.***> wrote: [https://avatars.githubusercontent.com/u/22754118?s=20&v=4]ncborcherding left a comment (BorchLab/scRepertoire#550)<#550 (comment)> Hey Anne, Thanks for continuing to follow up - I think you found a small bug: https://github.com/ncborcherding/scRepertoire/blob/040be3e1580700bd9d147f96c442c74c8bf545ea/R/combineExpression.R#L79-L81 Although combineExpression() is suppose to allow the frequency/proportion calculation by individual chains, the internal portion of the chain dissection called .offTheChain() has an internal filtering that was not not disabled here. Just pushed the relevent fix to the main repo - working on a larger bug in Seurat at the moment, but this should work for your purposes. Nick — Reply to this email directly, view it on GitHub<#550 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AIM7QPQTZEPSG6LEYAHHUVL33NV7PAVCNFSM6AAAAACKX7MPQ6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTIOJYGAZDANBVHA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[ncborcherding]

Hey Anne,

Yes you will with:

remotes::install_github("BorchLab/scRepertoire")

Make sure you restart your R session afterwards. I will get this fix up on bioconductor as soon as I can as well.

[annecar]

Thanks a lot! Will do. Anne Eugster, PhD Scientist, Research Group Binder Translational Immunology-Oncology Department of Biomedicine University Hospital Basel Hebelstrasse 20 4031 Basel Switzerland ***@***.*** On 6 Nov 2025, at 17:53, theHumanBorch ***@***.***> wrote: [https://avatars.githubusercontent.com/u/22754118?s=20&v=4]ncborcherding left a comment (BorchLab/scRepertoire#550)<#550 (comment)> Hey Anne, Yes you will with: remotes::install_github("BorchLab/scRepertoire") Make sure you restart your R session afterwards. I will get this fix up on bioconductor as soon as I can as well. — Reply to this email directly, view it on GitHub<#550 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AIM7QPUMMKGYLLRG2EQJYPT33N4HPAVCNFSM6AAAAACKX7MPQ6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTIOJYGMZDSMRZGM>. You are receiving this because you were mentioned.Message ID: ***@***.***>