Using genotype based demultiplexing tools on alevin-fry output

[changostraw]

Hello,
I am doing single nuclei RNA seq and have issues with cellranger discarding lots of reads and would like to directly compare between optimizing the cellranger reference to deal with overlaps etc. with the way that the alevin-fry pipeline deals with these issues and calling UMIs etc. However, my captures are multiplexed (by 4 samples), so after cellranger I demultiplex and de-doublet my barcodes with genotype based tools - demuxalot, demuxlet, souporcell, vireo etc. I run all these tools on my cellranger data and take a consensus, but I am not sure how to go about this step with the alevin-fry output as they all require a bam file to run and with alevin outputs there is a rad file instead of a bam file. I am not familiar with rad files and am not sure if they can be used instead of bam files or converted into bam files and I haven't been able to find this information. I am also concerned that since the pseudoalignment pipeline doesn't align to the genome I won't be able to use these genotype based tools to demultiplex. However, I also ran the selective-alignment pipeline with the --rad option, so perhaps those outputs are compatible with the genotype tools? Any suggestions would be helpful.

Thanks for the alevin-fry pipeline it ran very easily and fast!
cheers

[changostraw]

Actually, I am wondering if I can use the bam from cellranger since it contains all the barcodes and all the reads, and combine this with the filtered barcode list from alevin in order to run the demultiplexing tools. Does anyone see any issues with this?