About "No_contig"

[TsingYoo]

Dear Professor Kelvin,
I've used Dandelion for a few days, and discovered that this is a really powerful software. However, there is a question come during using Dandelion. I found that there were a large number of "No_contig" cells appeared not only when I read your documentation and examples, but also appeared when I processed my own data. But when I checked my data, I found that the match ratio of the VDJ from cellranger outputs is quite high compared with the large number of "No_contig" cells. So could you please tell me the reason of the conflict between "high match ratio of VDJ from cellranger outputs" and "low match ratio of VDJ from Dandelion"?
Thank you very much!

[zktuong]

Hi @TsingYoo, thank you for your interest in Dandelion. Can i ask you what do you mean by high match ratio of VDJ from cellranger outputs? Do you mean if you look at the barcode of a cellranger output file (e.g. filtered_contig_annotations.csv), you see that there are more cell barcodes with contigs?

Dandelion only adds a No_contig when the cell in the single-cell gene expression data does not match with any cell barcodes found in the VDJ data. This can arise from a number of reasons, but most likely due to incorrect barcode naming between GEX and VDJ data (especially if the cell barcodes are changed by the user).

For example, a common practice at sanger is GEX and VDJ library ids have separate ids, leading to a situation where the outputs have differen folder names, which can be confusing when adding the foldernames to the cell barcodes as a simple prefix.

[TsingYoo]

Thank you for your reply!

Your explanation of "high match ratio of VDJ from cellranger outputs" is exactly what I mean!
After reading your answer, I checked the barcode and found that the barcode is consistent with the Dandelion output file named "filtered_contig_dandelion.tsv", so I think the question might not caused by the unmatched barcode.
What's more, I read your tutorial documentation, and found that there are also a large proportion of cells with "No_contig" label from the sample files, such as the following image "https://sc-dandelion.readthedocs.io/en/latest/_images/notebooks_Q3-analysis_9_0.png". As you can see from the above picture, in the third UMAP image, "No_contig" cells possessed large area of the whole picture, just as what has happened to my own data.
I am very confused about it, is this an inevitable phenomenon?

[zktuong]

Hi,

That’s because most of the data aren’t T or B cells e.g. macrophages which don’t have TCR/BCR. Only T/B cells will get tagged with their contigs, otherwise it’s no contigs for all the other cell-types.

[TsingYoo]

Professor Kelvin,
Based on your guidance, I successfully solved the former question after processing my data again. Maybe I made some mistakes when running Dandelion from R which caused the question. Thank you again for your patience!
Besides, I noticed that there is a label named "Extra pair" in "chain_status" and a label named "TRB + Extra VJ" in "locus_status", can I ask you the specific meaning of "Extra pair" and "Extra VJ”?

[zktuong]

no worries!

Extra pair means whenever there is a possibility of more than 1 pair of chains e.g. TRB + first TRA or TRB + second TRA. That's also what the Extra VJ in the other column mean - there's more than 1 VJ (TRA) chain

[TsingYoo]

Do you mean that "Extra pair" and "Extra VJ" don't have exact significance?
We have some T cell samples from healthy controls and patients, and we found that there are more cells with the label "Extra pair" and "Extra VJ" in patients than in healthy controls. Could it caused by the illness, or just a coincidence?