huangyh09
commented
3 years ago Hi, thanks for sharing this issue.
What is the median or mean reads per cell in this experiment? Maybe you could also check the donor_ids.tsv file and see if it is caused by too few SNPs detected in these unassigned cells or the relatively low assignable probability prob_max.
If it is the latter, you could check the cumulative distribution function (CDF) of the prob_max and set a more lenient cutoff for assignable cells. By default, it requires prob_max > 0.9 to be assignable.
Alternatively, you could try the de novo mode by calling SNPs from the scRNA-seq directly with cellsnp-lite (mode 2b then mode 1a): https://cellsnp-lite.readthedocs.io/en/latest/manual.html
We are about to test these two strategies on low covered experiment and see they indeed help, but you can try it on your data too.
Yuanhua
Hi Huang, Thanks for reading my issue. I am trying to apply cellSNP and vireo on 10x single nucleus RNA seq. However, the unassigned nuclei are quite high.
The cellSNP run: cellSNP -s indexpossorted_genome_bam.bam -b barcodes.txt -O snpOUT_DIR -R genome1K.phase3.SNP_AF5e2.chr1toX.hg19.vcf.gz -p 20 --minMAF 0.1 --minCOUNT 20
The vireo run: vireo -c snpOUT_DIR -N 7 -o virosnp
Would you mind having a look and giving me some suggestion on how to improve the performance of cellSNP?
I hope I have provided enough information for you to help me with this issue. Thank you very much again.
Best regards,
Tongtong