10XGenomics / cellranger

10x Genomics Single Cell Analysis
https://www.10xgenomics.com/support/software/cell-ranger
Other
340 stars 91 forks source link

"Reference contains a contig longer than 536.8Mb (2^29 bp),Please split this contig." #148

Closed mihuzx closed 8 months ago

mihuzx commented 2 years ago

Hi,

Do you know how to solve this problem for huge genome? Or do you have any suggestions about how to split the genome and annotation at the same time?

Thanks Xuan

evolvedmicrobe commented 2 years ago

I am not sure which product or version produced this error, are you using cellranger arc?

mihuzx commented 2 years ago

Sorry for the confusion.

I was tried to build the index for wheat genome(17Gb) with cellranger-atac. and the code I used is cellranger-atac mkref --config=./wheat_iwgsc_v2.1_config_v2.txt

The error message is below:

Computing hash of genes GTF file...
...done

...done

mkref has FAILED
mkref has constructed an invalid reference, which is a bug.
Error: The supplied reference path /scratch/xz24199/GenomeRef/wheat_iwgsc/wheat_iwgsc_v2_cellranger_v2 is not a valid 10x reference on machine ra7-1.compute.lan:
Reference contains a contig longer than 536.8Mb (2^29 bp), which is not supported due to limitations of the .bai format. Please split this contig.
Please make sure that you have downloaded the correct 10x reference or generated a custom reference using `mkref`. Cell Ranger ATAC pipelines are compatible with Cell Ranger ATAC 2.0 or Cell Ranger ARC 1.0+ references. Cell Ranger ARC pipelines are compatible with a Cell Ranger ARC 1.0+ reference
Please contact support@10xgenomics.com.

Thank you for you time. Xuan

brettolsen commented 2 years ago

Apologies. This is a known issue with extremely large contigs caused by the limitations of our current indexing scheme. Your best path forward is to split the offending contigs to be below the limit. I would suggest splitting in regions with low mappability (e.g., centromeres) or far from annotated genes. I'm sorry, but we don't currently offer any tools to assist with this.