Does cellranger count require index fastqs

Question

[tdfyoder]

Hello,

Does cellranger count require index fastqs? The doc specifies the output directory of cellranger mkfastq (or csv options), however I have worked with the R1/R2 fastqs generated from another lab without issue. Thanks.

-Todd

Answer 1

[GenoMax]

You have likely seen this page on 10x support site. Sample index files are optional but you do need to have your datafiles named in a specific way (bcl2fastq). See section on My FASTQs are not named like any of the above examples at end of the page linked.

[tdfyoder]

Thank you, I do see the 'optional' notation now. I had concatenated R1/R2 fastq's from different lanes, in order to downsample, and had changed the catenated fastq to 'L001' to conform to the sampling nomenclature. I wasn't sure if this was problematic.

Answer 2

[geneticatt]

Some fastq files have index information in the read names instead of as separate files. Keeping indices in this format loses the quality score information associated with the indexes.

CellRanger doesn't use the index quality score, so I think using this format would be ok, but I don't know that the pipeline can read the indices out of the read name line of R1 and R2 fastq. That index files are listed as optional might imply that indices can be read from readnames, but there is nothing in the 10X documentation that says so explicitly.

[geneticatt]

I tested CellRanger using fastqs that have indices appended to read names and these were successfully demultiplexed into GEMs.