search

ncborcherding / scRepertoire

A toolkit for single-cell immune profiling
https://www.borch.dev/uploads/screpertoire/
MIT License
311 stars 54 forks source link

Phylogenetic tree #158

Closed mihem closed 1 year ago

mihem commented 2 years ago

Hi Nick,

have you thought about implementing phylogenetic trees? I think this could be quite useful in many situations. Ideally, I envision tracing clonal evolution using information from TCR/BCR plus the single cell object.

I just read about enclone, which creates phylogenetic trees with 10x data, but lacks the integration with single cell object and has a rather poor visualization I think (https://10xgenomics.github.io/enclone/pages/auto/tree.html). But maybe the output could be used for scRepertoire?

I also read that a similar extension was proposed in immunarch, but hasn't been implemented yet i think https://github.com/immunomind/immunarch/issues/150.

Thanks, Mischko

ncborcherding commented 2 years ago

Hey Mischko,

I think that is a good suggestion - I will add it to my to-do list. At least in the context of BCR, the approach would be able to map clonal lineages - I am not sure the utility of it in TCR - but it would be similar approach to TCRdist?

Also I think immunarch just released a dev version with this exact feature (I played around with it before the release and it was pretty fun!): https://github.com/immunomind/immunarch/blob/dev/R/phylip.R

mihem commented 2 years ago

That would be great. Also saw the following package just published which might be also helpful https://dowser.readthedocs.io/en/latest/

mihem commented 2 years ago

Hey Nick any news on that feature? Before it was just a fun idea of mine; now a reviewer specifically asked for that :D .

I know not the right place .. but I tried dowser and had problems with the format and in immunarch, I always got the error: No data in mandatory V.end column in sample airr_rearrangement! You said you ran that successfully with 10x VDJ data?

Thank you!

ncborcherding commented 2 years ago

Hey Mischko,

I have tried the implementation of something similar to immcantation/immunarch, but it requires a number of additional dependencies that would make the package flag on Bioconductor. Like the Kleinstein group's approach I think this is probably better as a stand-alone package, separate from scRepertoire. However, this might change if I can solve the issues related to dependencies.

I have gotten the immunarch pipeline to work on their example data, not 10x VDJ TCR data.

As of right now, I am still working on it. Sorry.

Nick

mihem commented 2 years ago

Hello Nick,

thanks for your swift response. Ah, too bad. But I understand that this implies quite some changes and additional dependencies.

Good luck with that, I am happy to do some early testing.

Best, Mischko