Filter clonotypes based on CDR3.aa

[luglilab]

❓ Questions and Help

We have a set of listed tutorials available on the website.

Hi,
Is there a way to filter clonotypes in an object based on CDR3.aa length — specifically to keep only those with a length less than 10 amino acids or greater than 20 amino acids?

Thanks!

[vadimnazarov]

Hi @luglilab thank you for using immunarch. Could you tell me the immunarch version you are using, please? To get the version, run the following code in R:

sessionInfo()$otherPkgs$immunarch$Version

If:

• it is 0.10.3
• your ImmunData object is called idata
• your sequences column called "CDR3.aa" (I would recommend renaming it to cdr3_aa), then you can do this:

# Filter and work with len < 10 & len > 20
idata_long_cdr <- idata |> filter(dd$length(CDR3.aa) < 10 | dd$length(CDR3.aa) > 20)

# Filter out (!) and work with len >= 10 & len <= 20
idata_filt <- idata |> filter(dd$length(CDR3.aa) >= 10, dd$length(CDR3.aa) <= 20)

dd is a special operator to call functions from DuckDB – the database immunarch uses on the backend via the immundata package. More on this operator here: https://duckplyr.tidyverse.org/articles/duckdb.html#call-arbitrary-functions-in-duckplyr. In short, you just call a DuckDB function name, like in the example above, I call the function length via dd$length – https://duckdb.org/docs/stable/sql/functions/text (find "length" on this page).

[luglilab]

Hi @vadimnazarov,

my version is 0.9.1 I didn't update yet my pipeline with the new version.

I have tried this command without success:

repFilter(
immunoarcobjFilter3,
.method = "by.clonotype",
.query = list(
nchar(CDR3.aa) >
morethan(10)
),
.match = "substring"
)

Simone

[vadimnazarov]

Hi Simone @luglilab, I'm sorry, but I no longer support earlier versions. Currently, I'm the only one who works on the tools, so it would require an enormous effort to support everything. If you provide a pipeline code or at least some parts of it, I can suggest how to migrate to the new immunarch version. A potential pitfall is that not all features are available yet, but I need to understand your code first to see if this could be a problem for your migration.