One BioSample per single cell ?

Question

[Charles Plessy]

In a single-cell analysis project we are about to deposit transcriptome data in the INSDC through DDBJ. Looking at prior art in NCBI's GEO, I see some examples where each single cell is registered as one BioSample. But I wonder if this defeats the purpose of BioSamples, that allow to link multiple records Barrett et al., 2012, whereas in such cases each BioSample is associated with a single data record only. In our case we have replicates batches of cell culture, and I wonder if it would make more sense to register these cell culture replicates as BioSamples, and link each single-cell transcriptome to their BioSample. What are Biostars' opinions ?

[fanli.gcb]

I thought GEO handles the BioProject and BioSample creation part? At least that's why I prefer GEO >>> SRA