Forum:BCL files as deliverables from a sequencing center

Question

[Ido Tamir]

Hello,

Recently my customers started to request BCL files from our facility. Is this something sequencing providers or core facilities usually give their customers? Is it a default option or only under some conditions (i.e is there a surcharge in addition to the normal output of FASTQ files)?

Thank you very much for your input.

Ido

Answer 1

[Devon Ryan]

No, this is incredibly unusual. What is likely happening is that your clients are using CellRanger and think that they need to have it do demultiplexing for them. They don't need to do that, since it's perfectly happy to start with fastq files. There are of course issues with giving them BCL files if the relevant lanes have pools with samples from other clients and this is why I have never given any group BCL files. At the end of the day, I have yet to see a group actually needing these files, they just think they do from the cellranger documentation.

[Ido Tamir]

Thank you very much for your answer. Yes, a part of it is driven by CellRanger.

[GenoMax]

We find it useful to run the demux using 10x software since it produces a folder per sample that contains the demultiplexed files (with index reads in a separate file) as needed for downstream analysis using 10x software. There is also a subtle difference (see below).

While it is true that 10x genomics software uses a wrapper for bcl2fastq here was what 10x support had to say about using proper variant of 10x software (cellranger or longranger mkfastq) when doing the demultiplexing. Note: fastq files produced should be identical irrespective of the software used.

However, the qc statistics output by mkfastq will be different even if the output FASTQs are the same. For example, barcode_exact_match_ratio and bc_on_whitelist. Since Long Ranger and Cell Ranger are using different barcode whitelists, this is where the wrapper portion of mkfastq diverges between pipelines.

Looking to the future, there could be changes to future versions of the mkfastq pipelines that cause them to diverge. So as a general best practice we recommend sticking with the mkfastq version for that specific product. (This is already true for cellranger-atac mkfastq which has diverged from the others because ATAC libraries are dual indexed).

Answer 2

[Raony Guimarães]

Is this something sequencing providers or core facilities usually give their customers?

No, this is not very common but I can understand why someone would ask for that. (i.e. They prefer to do sample demultiplexing themselves adjusting the penalty parameters or use a different software rather than bcl2fastq).

Is it a default option or only under some conditions (i.e is there a surcharge in addition to the normal output of FASTQ files)?

Only under some conditions, BCL files are much larger than FASTQs so there should be a surcharge to send them back. It will be a lot more work to deal with these large files for sure.

[Ido Tamir]

Thank you very much for your answer. We always also offer the complete lane, so they can always demutliplex the data themselves.