Clarification Needed on Dataset Processing

[nicole1q]

Hi， @M0hammadL， @b-schubert， @ArcaneEmergence， @irene-bonapa and @drEast

I am currently working to reproduce your meaningful study, "Multi-modal generative modeling for joint analysis of single-cell T cell receptor and gene expression data." However, I have encountered some issues while processing the datasets and would greatly appreciate your guidance.

1. Regarding the Borcherding Dataset
   I recently downloaded the Utility v0.0.4 dataset from Zenodo, but I couldn’t locate the rna_combined.h5ad and tcrs_combined.csv files mentioned in preprocessing/Borcherding_preprocessing.ipynb.

When exploring the dataset:

The h5ad files in utility/data/processedData/individualSeurat/h5ad/ contain only gene expression (GEX) information.
The TCR annotations in utility/data/processedData/combinedDataSets/ use barcodes (e.g., barcode) that do not match those in the h5ad objects (e.g., BCT1.1_AAACCTGCAGATCGGA-1).
Could you clarify where to find the paired h5ad and TCR annotation files?

2. Regarding the COVID Dataset
   I downloaded the dataset from GSE171037, but encountered similar issues:

The file GSE171037_TcellReversePT_integrated_Tcells.h5ad only contains GEX data.
The raw data in GSE171037_RAW includes only orphan TCR information (e.g., 'orphan VDJ', 'orphan VJ') without paired TCR data.
Additionally, I couldn’t find the Covid19_TAs_PBMCs.h5ad file mentioned in your work.

Could you please provide guidance on where to find the complete datasets or suggest an alternative approach?

Thank you very much for your time and assistance.

[drEast]

Dear Nicole,
Thanks for your interest in our study.
Unfortunately, we faced a bit of a reproducibility issue here.

For uTILity: We downloaded the data when it was still unversionized. Back then, I downloaded the data from a Google Drive folder (which does not exist anymore) linked in the repository and converted it myself from Seurat to h5ad.
I see two options here:

1. You process the current version of uTILity to the data format outlined in the tutorials and re-use mvTCR on the new data. The results however will not be comparable, as uTILity has increased since then.
2. We kindly ask Nick Borcherding whether I can share the data with you over a dropbox. The current license does not allow for derivatives as far as I see. So I would need Nick's OK on this one. Let me know if I should contact him.

For Covid:
Please try the file "GSE171037_TcellReversePT_munich.h5ad". I double-checked, it has all the cells in it that we used in our tests. The Chicago cohort did not have any TCR information, so this information was removed from the joint data object as well as from mvTCR.

I hope that helps. Please let me know if there is anything else.
Best,
Felix

[nicole1q]

Dear Felix @drEast ,

Thank you for your detailed explanation and guidance regarding the data reproducibility issues. I truly appreciate your efforts in clarifying the situation.

I have already contacted Nick Borcherding for assistance regarding the Utility v0.0.4 dataset from Zenodo.

Would it be possible to share the data with me via a link? My email address is qixiaoning1217@gmail.com. I understand the licensing constraints and would be happy to seek Nick Borcherding's approval if necessary.

Looking forward to your response. Please let me know if there's anything else I can do to facilitate this process.

Best regards,
Nicole