11 weeks ago by
Cambridge, United Kingdom
Because they involve different protocols? Bulk RNA-seq tends to use (ribosome-depleted) total RNA protocols nowadays, while most single-cell RNA-seq uses poly-A'd approaches. I can imagine that this would result in different biases and preferences for particular transcripts.
There are also considerations with cell dissociation and size. For example, if a tissue contains some fragile cell types, these would lyse and not show up in the single-cell data. In comparison, the fragile cell types would still be present in the bulk data where no dissociation is required, only lysis of the entire tissue. The resulting bulk-only transcripts would compete with and suppress the coverage of transcripts unique to other cell types, resulting in counts that are only observed in single-cell data. A similar effect occurs with large and small cells in the same bulk population, where transcripts unique to small cells get suppressed in bulk data.
Finally, there is always sampling noise, which means that transcripts for lowly expressed genes may be sampled in a few cells on a plate but not in the bulk sample. You would have to have equal total sequencing depth between the bulk sample and all single-cell samples for the counts to be fully comparable. Obviously you will miss transcripts if the bulk sample is not sequenced to the same depth.
P.S. This question is more suited for a general forum like SEQAnswers, it doesn't seem to involve any Bioconductor packages.
modified 11 weeks ago
11 weeks ago by
Aaron Lun • 17k