Renaming is not completed when loading MIXCR data into immundata using read_repertoires()

[eliasmrmueller-max]

🐛 Bug

When loading my MIXCR .tsv files using read_repertoires() I keep getting the error
"Error in agg_receptors():
! Not all columns in the receptor schema present in the data: [cdr3]. Please double check and run again."
I have tried using cdr3_aa, v_call and Immunarch 0.9.1 names such as CDR3.aa and V.name. I always got the same error.
Upon trying with a MIXCR column name (e.g. CloneId) there are no errors.
Idata$annotations$ shows the names of MIXCR columns, not immunarch/Immundata columns.

When I use the repload -> from_imunarch() method I do not have this problem.
I would keep doing it this way, however when using this method my metadata is not integrated into idata$repertoire.

Code (basically the tutorial code)
md_file <- file.path(TSV_DIR, "metadata.txt")
md_table <- read_metadata(md_file)
tsv_files <- list.files(TSV_DIR, pattern = "\.tsv$", full.names = TRUE)
idata <- read_repertoires(path = tsv_files,
schema = c("cloneId"),
metadata = md_table,
preprocess = make_default_preprocessing("airr"),
repertoire_schema = "Donor")

Printing: idata$annotations
Results: annotations

A duckplyr data frame: 52 variables

cloneId readCount readFraction uniqueMoleculeCount uniqueMoleculeFraction
.......
47 more variables: targetSequences , targetQualities ,

allVHitsWithScore , allDHitsWithScore , allJHitsWithScore ,

allCHitsWithScore , allVAlignments , allDAlignments ,

allJAlignments , allCAlignments , nSeqFR1 ,

minQualFR1 , nSeqCDR1 , minQualCDR1 , nSeqFR2 ,

minQualFR2 , nSeqCDR2 , minQualCDR2 , nSeqFR3 , …

-> All of these are MIXCR column names

Thanks in advance!

[eliasmrmueller-max]

I just sent a set of the MIXCR output along with an extract of the metadata to support@immunomind.com

[vadimnazarov]

Hello @eliasmrmueller-max thank you for using immunarch and sending the files, it's very helpful. I believe what you encountered is not a bug but expected behaviour. To read MiXCR files into R with immunarch 0.10.3 (and hopefully later), you can do this, and I use the files you sent me as an example:

md_table <- read_metadata("~/Downloads/Github_issue/metadata.txt")
md_table

idata <- read_repertoires(
  path = "~/Downloads/Github_issue/*.tsv",
  metadata = md_table, 
  schema = c("cdr3_aa", "locus"),
  count_col = "readCount",
  rename_columns = c(cdr3_aa = "aaSeqCDR3")
)

First, we read metadata. Mind that the "File" column could contain both the full path like "~/project/my_project0123/data/filename1.tsv" or just a filename "filename1.tsv" if "metadata" is in the same folder as files.

Second, mind the schema and rename_columns. Immundata first renames columns from aaSeqCDR3 to cdr3_aa, and then creates receptors using the provided schema.

But I would recommend using https://mixcr.com/mixcr/reference/mixcr-exportAirr/ for MiXCR to get an AIRR-C compliant output. It will make the further analysis (and development for me...) much, much easier. Just to give you an example: there is no "best VDJ gene segment" per receptor in the MiXCR data you sent, only "all VDJ hits", which is not used and won't be used by immunarch. Either outputting a more detailed MiXCR format file or using exportAIRR could fix this issue.

[eliasmrmueller-max]

Thank you so much for the fast (and helpful) response!