huangyh09
commented
3 years ago Hi, in principle you can use vireo to separate human and mouse cells after mapping reads to a combined genome with concatenating human and mouse genomes, while there are difficulties to use vireo in practice.
First, it is difficult to compile a VCF containing both mouse and human, as they don't share chromosomes. Second, even using vireo's de-novo setting, one could genotype SNPs across the combined genome, but the main difference comes from read coverage instead allele frequency, which is not what Vireo's aim.
As you mentioned there are much easier solutions. If you align them to the combined genome, you can simply count the expression level of a few human genes and mouse genes, which will also indicate the cross-species doublets (if one barcode has both human and mouse genes). This should also work if you perform clustering with human and mouse genes.
Yuanhua
Hi, I am wondering if I can use cellSNP+vireo to demultiplex mouse+human mixed barcodes, i.e. the barcodes which have both human and mouse cells (doublets). If so, do I provide the bam file which is aligned using a chimeric mouse+human genome?
I know this can be done using mapping each cell to the mouse and human genome individually, but I would like to try it based on the SNPs.
Although,CellSNP+vireo works very well to identify barcodes from two different human donors based on their genotypes (which makes sense), but also, relating to that, is it possible to even find homotypic doublets/multiplets in the data- since a barcode with two cells from the same human (for example) should have a slightly different genotype probability than one human cell? (Not sure)