Idea for scSPLASH

[dannyconrad]

Really cool work, and I managed to get scSPLASH working on the first try which is always nice.

I had a thought for scSPLASH, could the same statistical framework be used to detect differential gene expression (not just editing/splicing/sequence diversity) if the cell barcode is used as the anchor instead of kmers? Unclear how exactly to define the targets given the R2 reads do not all start at the same position along the transcript.

If this could be implemented, the idea would be that you could use scSPLASH to help dentify differential expression of unexpected or poorly annotated transcripts, such as transposable elements, rRNAs, viral genes, etc.

Would love to hear your thoughts @marekkokot @roozbehdn

[roozbehdn]

Hi Danny,

Glad to hear that you have been able to run scSPLASH. Yes, in theory, sc-SPLASH could be used for differential gene expression analysis. By using much shorter anchors than the default length (k = 27), the anchor sequence would likely occur in many places across the genome, and the different targets identified by sc-SPLASH for that short anchor could then correspond to different genes. We have not yet tested this idea systematically though. And yes I agree with you, this would have the potential to detect differential gene expression for poorly/not annotated genes/transcripts.