drtconway
commented
5 years ago I don't have much experience with single cell data (genomic or RNASeq), but it sounds to me like there could be some step in the library preparation that biases toward human reads. For example, if one used a standard exome protocol on xenograft material, you'd expect a bias in favour of human reads.
A couple of things you might like to check:
Does the 3' protocol just pull down on the polyA tail?
Is there some selective amplification step?
Xenome itself is very simple, so it seems to me a bit unlikely that it is directly causing problems, but it might be an issue either with the reference or the protocol.
Tom.
Hi,
I have previously used Xenome to classify mouse reads from human reads from bulk PE RNA-seq data and the tool has worked very well. Now I would like to apply the same technique for scRNA-seq data (10x 3’ scRNA-seq, where R2 represents cDNA), but I have run into problems. It seems that Xenome cannot distinguish mouse reads from human reads, and prints out roughly the same percentage of reads classified as “human” across samples that include human-only and human-mouse samples.
So, can I apply this tool for 3’ scRNA-seq data? Am I doing something wrong?
Here is the code I have been using: xenome classify -T 8 -P /path/to/reference -i /path/to/R2.fastq —graft-name “human” —host-name “mouse” —output-filename-prefix “test”
Thanks.