Spliced and unspliced counts from HIVE scRNA-seq data

Question

[anna_o]

Hi!

I was wondering if anyone has successfully performed RNA velocity analysis on scRNA-seq data generated using the HIVE CLX technology from Honeycomb.

Specifically, I am trying to figure out how to generate sliced and unspliced count matrices from this data.

The technology does not use UMI's for duplicates' identification. Instead, due to the nature of the library preparation, it is possible to count molecules without UMI's because reads have random alignment positions, so the alignment positions + cell barcodes are used to identify PCR duplicates. Their bionformatic support suggested I deduplicate the bams with Picard in a cell barcode-aware way and use these as an input to velocyto. However, when trying to run velocyto on these, I end up raising a following error (this happens even, when I try with the -U flag for the UMI-less data):

raise IOError("The bam file does not contain cell and umi barcodes appropriatelly formatted. If you are runnin UMI-less data you should use the -U flag.")

The files do contain the barcode flag CB. Any advice or suggestions for how to navigate this would be appreciated, especially if anyone has managed to run this type of analysis on HIVE data before!

Thanks!

Answer 1

[predeus]

Hi Anna,

I think you can add a fake "corrected UMI" sequence (UB) to your Picard-deduplicated BAM file. Basically adding a 10nt random string should work.

Hope this helps!

-- Alex

[anna_o]

Hi Alex,

Thanks a lot! I will definitely try this. Anna