Paired OAS Help

About paired OAS

Sequencing of paired BCR repertoires is known to be a hard task due to the need to physically link or label natively paired variable heavy (VH) and light (VL) chains. With the advent of 10xGenomics sequencing, full-length natively paired VH/VL sequences can now be obtained, although at the expense of the B-cell throughput number in comparison to unpaired Illumina sequencing.

The Observed Antibody Space (OAS) database now provides access to annotated paired sequences from 10xGenomics studies. To date, OAS collates over 120,000 paired sequences from 7 different studies. The data is available for download or you can filter the sequences with respect to certain metadata parameters using our search form. To download the data go to the Search page.

Paired sequences in OAS can be filtered according to attributes such as species, disease, vaccination etc. The fields are non-exclusive, meaning that the user could choose a combination of fields that does not exist in our database.

Metadata

Similarly to the unpaired version of OAS, all datasets are organized into studies, that are in turn subdivided into data-units. A single data-unit is a set of sequences uniquely identified by its metadata. The range of meta-parameters are:

• Age Information on age of the human B-cell donors.
• Disease Indicates whether the donor was sick at the time of B-cell extraction.
• Vaccine Indicates whether the B-cell donor was purposely immunised prior to B-cell extraction.
• BType Indicates whether B-cells were experimentally sorted prior to sequencing.
• Species Organism of the B-cell donor.
• Author First author and the publication year.
• Unique sequences Number of sequences that passed quality filtering steps.
• BSource Which organ/tissue the B-cells were extracted from.
• Subject Indicates whether the B-cells can be tracked back to a particular individual.
• Longitudinal If the study is conducted over a period of time, indicates the particular timepoint when B-cells were sourced.

Data Format

A series of .csv.gz files will then be downloaded to your current directory. Each .csv.gz file contains as the first line the metadata for the data-unit and the following lines each sequence and its annotations.

The contents of each data-unit file can look as below:

Linking heavy and light chains

In the ideal case scenario, 10xGenomics sequencing would yield 1-to-1 heavy/light chain pairings for each interrogated B-cell. However, in many cases more than one heavy and/or light chain sequences harbour identical 10xGenomics cell barcodes. Linking such heavy and light chain sequences can lead to combinatorial inflation of the real sequence number and incorrect estimation of the repertoire diversity (Figure 2).

One solution is to filter out sequences whose 10xGenomics barcodes are shared between more than one unique heavy and light V(D)J recombination events. This step has already been performed for each data unit in OAS. However, a recent study showed that individual B-cells can produce more than two functional V(D)J recombinations (Shi et al., 2019. Cell Discovery).

Contact

If you would like to contact us or to submit your study to OAS, please drop an email to oas_opig@stats.ox.ac.uk.