Question: Can I perform shallow sequencing to assess the quality of Single Cell 3' Gene Expression libraries?
Answer: Shallow sequencing can be performed to obtain an initial assessment of library quality prior to performing deeper sequencing. This Illumina Technical Note provides a demonstration of shallow sequencing on the Illumina iSeq to re-balance library pools prior to deeper sequencing. While iSeq can be used for pool rebalancing, assessing library quality should be performed on one of our validated sequencers such as the MiSeq.
Metrics including "Valid Barcodes", "Valid UMIs" and "Reads Mapped Confidently to Transcriptome" are accurate even at low sequencing depths.
However, the "Estimated Number of Cells" and "Fraction Reads in Cells" may not be accurate at low sequencing depths. The sequencing depth needed for accurate cell calling is highly variable and depends on the complexity of the sample.
When working with PBMC samples we have found that as few as 500 read pairs per cell can be sufficient for cell calling, as described in our Technical Note.
However, when working with cells with higher RNA content (eg. cell lines), more complex or heterogeneous tissue types (eg. dissociated tumor cells), or cells with lower viability and poorer sample quality, we have found that you may need more than 10,000 read pairs per cell for accurate cell calling.
The figures below show an example of a complex dissociated tumor cell dataset that was sequenced deeply and then downsampled to lower depths.
Figure 1: Cell calling
- Cell Ranger called ~2,500 cells when the library was sequenced to a depth of 40,000 read pairs per cell
- Cell Ranger called ~1,000 cells when the library was downsampled to 500, 1,000, or 2,000 read pairs per cell
- Therefore, there was a 2.5-fold difference in the number of cells called at low and high sequencing depths
Figure 2: Fraction Reads in Cells
-
- Fraction Reads in Cells was ~94% when the library was sequenced to a depth of 40,000 read pairs per cell.
- Fraction Reads in Cells was ~80% when the library was downsampled to 500, 1,000, or 2,000 read pairs per cell
- Therefore, there was a 14% difference in the Fractions Reads in Cells at low and high sequencing depths
Figure 1
Figure 2
Products: Single Cell Gene Expression