10Xv2 scRNA dataset SRR files return only R1, how to run cellranger ?

Question

[jo.bac]

I am trying to run scRNA velocity with velocyto. A preliminary step is to run cellranger on fastq files but I have an issue with my chosen dataset

Here is what I did:

• Downloaded each SRR from GSE104323 using a loop with fastq-dump --split-files --origfmt --gzip SRR6084134
• This only yielded one SRR file instead of 3 expected files (R1,R2,I1)
• I renamed my fastqs to:

SRR_S0_L001_R1_001.fastq.gz SRR_S1_L001_R1_001.fastq.gz ...

Then tried to run:

cellranger count --id=test --fastqs=fastqgz/ --transcriptome=refdata-gex-mm10-2020-A

which failed with error:

The read lengths are incompatible with all the chemistries for Sample SRR in "/mnt/c/Users/jobac/Downloads/SRA_split_files/GSE104323".
 - read1 median length = 98
 - read2 median length = 0
 - index1 median length = 0

I suppose the problem is that I have only one file instead of separate R1,R2,I1. How to obtain them for this dataset or work around this issue ?

Thanks a lot for the help!

Answer 1

[GenoMax]

As best as I can tell this submitter has submitted these samples in a strange format where the cell barcodes and UMI appear to have been moved to the header of these R1 fastq files. It is going to be a pain to deal with this data since it will likely require some custom coding to get it back into R1,I1,R2 format. Here is an example of one read.

@K00110:126:HJJTVBBXX:4:1101:20151:5464_CACGGATGGG_CAGTGCATGGATGG_
ACAAGGACGGGATAAAGTCCGAGAAATGTTCATGAAGAATGCCCATGTCACAGACCCCAGAGTGGTTGATCTGCTGGTCATTAAGGGAAAGATGGAGC
+K00110:126:HJJTVBBXX:4:1101:20151:5464_CACGGATGGG_CAGTGCATGGATGG_
AAFFFJJJFJJFJJJFJJJJJJJAFFJJJJJJJJJFJJJJJJJJJJJJJJJJJJJJJFJJFJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJFJJ

On top there are hundreds of such samples.

If you are just trying to use this as a test then find some other dataset with proper R1,I1,R2 files.

[jo.bac]

Thanks a lot ! I looked in detail at one example and found this comment by the authors:

"The Unique Molecular Identifier and cellular barcode corresponding to each read have been appended to the read id separated by an underscore."

Unfortunately I need to analyze this data specifically. It is my first time dealing with fastq, I am a bit at a loss how to split the files back. If someone finds time to do one example highlighting text and showing which parts to split back into the three files I could write a custom script

[GenoMax]

v2 barcodes should be 16 bp and UMI's will be 10 bp. It appears that the barcodes in the example above may have only been sequenced as 14 bp.

So from

@K00110:126:HJJTVBBXX:4:1101:20151:5464_CACGGATGGG_CAGTGCATGGATGG_

CAGTGCATGGATGG - cell barcode
CACGGATGGG - UMI

Illumina index is likely gone for good. You could randomly use one of the valid Illumina indexes to create a fake I1 file.