Unable to retrieve spliced/unspliced RNA from single nuclei isolation

[hyjforesight]

Hello STARsolo, Thank you for making such an amazing packge!

We’re unable to retrieve spliced/unspliced RNA from single nuclei isolation procedure of scRNA-seq Flex kit. Please find the attached figure.

ACI sample is used as control. We do single cell isolation of tumor, live cell sorting, then use 3’ chemistry v3.1 kit to do the scRNA-seq. We use CellRanger v8.0.1 to convert Fastq to matrices. Then, use velocyto.py v0.17 to retrieve spliced/unspliced RNA (generate loom file) from CellRanger outputs. We can see 7000 n_genes_by_counts for matrices and 5000 n_genes_by_counts for loom file, which is OK for our RNA velocity analysis. These results validate that these procedures are good to retrieve spliced/unspliced RNA from single cell isolation.

1503095E, 2113752H, 2300648F, 2121949E samples are FFPE samples. We do hydration, dissociation, single nuclei isolation, add probes embedding for multiplexing, then use Flex kit to do the snRNA-seq. We use CellRanger v8.0.1 to convert Fastq to matrices. Then, use velocyto.py v0.17 to retrieve spliced/unspliced RNA (generate loom file) from CellRanger per_sample_outs. We can see n_genes_by_counts of loom file is extreme lower than matrices, which cannot be used for our RNA velocity analysis.

For this situation (unable to retrieve spliced/unspliced RNA for our FFPE samples), could you please help with our questions:

1. Is velocyto.py not compatible with the single nuclei isolation? Both Fastq files from v3.1 kit and Flex kit are compatible for Cell Ranger, so I don't think it is the kit problem. The splicing of mRNA happens in nucleus, so single nuclei isolation should retrieve good spliced/unspliced RNA, isn’t it?
2. Did your amazing STARsolo package met similar issue or can it retrieve the right spliced/unspliced RNAs?

Thank you! Best,

[apredeus]

Hi, This is probably not the right place to ask such detailed questions I'm afraid. However I can briefly comment - hopefully it would be useful to you!

Yes, velocyto is normally compatible with snRNA-seq, we have ran it on such data. I am not sure why it didn't work in your case. STARsolo is capable of generating the same matrices and should work as well. However, seeing that you have FFPE samples makes me think it might be a probe-based protocol? If that is so, you would not be able to get the spliced-unspliced information, I'm afraid.

Hope this helps

-- Alex

[hyjforesight]

Thank you Alex! You're right. It's the problem of Flex kit. See this: https://kb.10xgenomics.com/hc/en-us/articles/25938615598477-Is-trajectory-analysis-compatible-with-the-Fixed-RNA-Profiling-Flex-assay