Using AIRR Community Reference Sets with 10X Cell Ranger

The download_germline_set utility can be used to download germline reference sets from the AIRR-C OGRDB database in the format required by Cell Ranger’s makevdjref tool.

OGRDB currently only supports 10X format for human IG sequences. Support for Rhesus macaque IG sequences is planned for a future release.

Example Usage

In this example, we follow the 10X Cellranger tutorial, but modify the reference set to use the latest version of the human IGH germline set from OGRDB, downloaded with the download_germline_set utility. It is assumed below that you have used the directory layout described in the tutorial.

Prerequisites

First follow the instructions in the tutorial Running Cell Ranger VDJ. Complete it, so that you have the sample data annotated with the build-in Cell Ranger data set. Open the file runs/HumanB_Cell/outs/airr_rearrangement.tsv. Note that the annotations are at gene level: the 10X reference set does not contain allele-level information.

Annotation with OGRDB germline reference set

In the top-level working directory, create a new directory ogrdb_ref. cd to the directory and enter the command:

download_germline_set "Homo sapiens" IGH -f 10X

The tool will download germline reference data for all IG loci (even though only IGH is specified) and will create a single FASTA file. You may see warnings that some sequences have been omitted because no leader sequence is available. Cell Ranger requires a leader sequence for annotation, and there are a small number of records in the OGRDB human IG set which do not at present have an annotated leader: these will not be included in the 10x reference.

Now enter the command:

cellranger mkvdjref --seqs Homo_sapiens_IG_10x.fasta --genome ogrdb_mkvdjref

Here, –seqs specifies the input FASTA file containing the reference sequences from ogrdb, and –genome specifies the name of the output directory where the new Cell Ranger reference will be created.

Now cd to the runs directory, and create a new run that uses the OGRDB reference set:

cellranger vdj --id=HumanB_Cell_OGRDB \
    --reference=../ogrdb_ref/ogrdb_mkvdjref \
    --fastqs=../dataset-vdj-practice/sc5p_v2_hs_B_1k_multi_5gex_b_fastqs/sc5p_v2_hs_B_1k_b_fastqs \
    --sample=sc5p_v2_hs_B_1k_b \
    --localcores=8 \
    --localmem=64 \

When Cell Ranger completes, open the file runs/HumanB_Cell_OGRDB/outs/airr_rearrangement.tsv and note that the records are annotated at allele level.

Thanks to Eve Richardson for the investigation, core code and testing, and to Katherine Jackson for helpful support.