Extract C-region sequence

[GeroKn]

Hi all,

I have RNAseq data of murine B cell lymphomas and I want to analyze the clonality of these lesions. I'm running this command:

mixcr analyze rna-seq --species mmu --assemble-longest-contigs

followed by

mixcr exportClones

I get a table with target sequences, V, D, J hits and also C hits. However when I blast the target sequence on https://www.ncbi.nlm.nih.gov/igblast/, there is nothing that maps to the C region, only CDR3.

What I would like to receive is the longest possible contig, including the fragment that allows the algorithm to call the C region. I'm figuring that the sequence must exist, otherwise there wouldn't be a C hit. How can I extract it?

It would be great if someone could help me out, thank you very much in advance!
Gero

[tatitrav]

I have the same question and am wondering if you ever figured it out. I have been able to find a C hit on my own with the target sequences, but I have several samples that cut off the target sequence right at the end of the J.

[mizraelson]

Hi everyone,

MiXCR does use the C gene sequence, if present, to identify the best C gene hit. However, it currently can’t be used in the assembling feature, so the longest sequence you can assemble clones by starts with the 5’UTR and ends with FR4.

@GeroKn There is no need to run exportClones as mixcr analyze already includes it. Since you’re working with non-enriched RNA-seq data, MiXCR tries to assemble the longest possible contigs, but the most important part is the CDR3 because it defines the clone. For this type of data it is common to get only a few clones, and if the reads are short (which is typical for bulk RNA-seq), you will likely only recover the CDR3 sequence.

@tatitrav In RNA-seq preset you might have multiple target sequences and indeed if one read was covering both FR4 and a part of the C gene you will see it there, but MiXCR does not cut it off, it may be that for some clones there were no such reads. What command did you use? Could you share an example output file?

[tatitrav]

@mizraelson Thank you for the response! So are you saying the sequences associated with the C gene are not typically included in the target sequence for each clone?

I ran the standard command using an imported library for canine TCR/BCR genes:
mixcr analyze rna-seq
Here's an example IGH output file for one of my samples. CI154629.clones_IGH.tsv I see that I have top C hits for most of the top clones. I guess my question is similar to @GeroKn's. When I look at the target sequences for the top two clones (#1 and 4) and align those to germline sequences for canine VDJs, I see that clone 1's target sequence ends right at the J. However, clone 4 has a longer target sequencing that includes some of the constant gene. My original thought was that the abundance of reads would affect the sequence output, but with clone 1 being more abundant than 4, I would expect more reads and therefore more coverage.

Do you have any additional insight or recommendations to potentially get a longer target sequence that would include the C reads?

Thanks in advance for your help!

[mizraelson]

For all non-fragmented libraries C gene is never present in the target sequence. For fragmented data like RNA-seq it can be in the target sequences, but this is rather a byproduct of the algorithm for assembling repertoires in such cases. Unfortunately we currently do not have any way to assemble the C gene.