Open LTLA opened 4 years ago
Thank you very much ! I think I can try to follow scRNAseq package tutorial to prepare my data first if you think that is OK. I would send another mail to you ( attached my email address ) for details.
I think I can try to follow scRNAseq package tutorial to prepare my data first if you think that is OK.
I suggest having a look at the make-...-data.Rmd
and make-...-metadata.R
files in the inst/scripts
subdirectory. You can just pick one of them - aztekin
for an easy example, segerstolpe
for an epic example - and adapt it to your data.
I would send another mail to you ( attached my email address ) for details.
I don't mind email, but unless there's something sensitive, I'd prefer to do the discussion here. This makes it easier to keep track of decisions and allows other interested people to contribute.
If you must use email, use one of my maintainer addresses, e.g., from scran.
@ahy1221: I'll apologize for not answering your email directly; my old CRUK account forwards my mail but does not forward the sender address, leaving me unable to respond directly to anyone who emails me. So here we are.
Anyway, regarding your data: yes, it would be great to put it in here. The advice in the vignette is probably a bit more discouraging than I intended. What I wanted to say was that 10X data is very standardized already, so it's not very hard to read in compared to, say, the Zeisel dataset (which, BTW, was very painful to parse); there's not much end-user value to be added from hosting 10X datasets here, but if you're willing to set up scripts and metadata, I'm willing to put them in.
I should mention, though, that I don't intend this package to be the "source of truth" for any data. This means that you still need to have the original count matrices hosted on some recognized repository like GEO or ArrayExpress (which you should have, anyway, because you need to upload the FASTQ files uploaded somewhere). scRNAseq simply provides a convenient interface for R end-users to access the data; if we were to be a source of truth, we would have to be a lot more careful with data provenance, and I don't think we're set up for that.