dcjones
Xenium 5K Proseg2 output questions
Hi,
Thank you for all your work on Proseg. I ran Proseg2 on Xenium 5K and I have few questions please:
I get a warning on the Xenium analysis summary :
Is there a way to eliminate those from the input
My understanding is that you use the prior DAPI segmentation info, the outputs in the Xenium explorer do not include the nuclear segmentation, is there a way to show the nuclear segmentation and the Proseg segmentation, and retain the nuclear trasncripts info
Are the multimodal segmentation (interior RNA stain and boundary stains) included or will be included in Proseg algorithm?
You should be able to filter these transcripts out with --min-qv 20. I should probably make these the default, but it's typically a very small number of transcripts.
Right now proseg doesn't output or retain any information on the original nuclear segmentation, but it would not be hard to add a column to the output to account for that.
There is an option in proseg now to penalize deviation from the full cell segmentation, which might let it better take account of boundary stains. This is disabled by default because I haven't tested it much, so can't say for sure what benefit there is, if any. If you want you can try e.g. --prior-seg-reassignment-prob 0.3. Hopefully in the future I'll have benchmarked this more, and have more clear guidance on how to use it.
