How to Add a Tag to UMIs to Process Libraries from a Single Individual Together?

[marcoschinas]

I am processing multiple libraries from a single individual. My goal is to prevent conflicts where the same UMI might appear in different libraries. I want to add a unique tag to the UMIs for each library before processing them together.

Will the following example code handle the issue?

`filename=BE01_{{UMI:n}}EKDL240033176-1A_22FNKJLT4_L{{n}}{{R}}.fq.gz

mixcr analyze takara-human-rna-tcr-umi-smarter-v2
--species hsa
$filename
"~/mixcr/BE01_results"
--assemble-clonotypes-by CDR3
`

[mizraelson]

By default, MiXCR allows and resolves cases where several clones are marked with the same UMI. Could you try running it without adding an extra tag and share the assembly report? From the report, we’ll be able to determine if there’s an issue or not.

[marcoschinas]

Here is the assembly report for the run without adding an extra tag,

Overall, both approaches give me a similar amount of clonotypes. I also ran each library independently, when I merged these individual files and I compared them with both approaches, I obtained a higher overlap on the total amount of unique CDR3.nt sequences by adding the UMI tag. In both cases, I obtained around 100 new clonotypes compared to processing each file independently.

BE01_01_EKDL240033176-1A_22FNKJLT4_results_bysubject.assemble.report.txt

[mizraelson]

Number of pre-clonotypes per group:  
    0: + 1301 (1.39%) = 1301 (1.39%)
    1: + 90927 (97.39%) = 92228 (98.79%)
    2: + 1103 (1.18%) = 93331 (99.97%)
    3: + 31 (0.03%) = 93362 (100%)

Here, we see that for 97.39% of UMIs, a single consensus was found. Then there is 1.18% where two consensuses were found. The result depends on their ratio, but in most cases, MiXCR resolved them as two distinct clonotypes.

Overall, I would say there were no significant issues with UMI collisions.