Question: Can ERCC spike-ins be used for normalization?
Answer: It is technically feasible, but not recommended.
We do not recommend the use of ERCC controls to asses sample-to-sample variability when generating 10x Single Cell libraries for the following reasons:
- The ERCC RNA spike-in mix contains exogenous transcripts that differ in length and GC content when compared to the endogenous human transcriptome.
- Because the exogenous RNA will be evenly distributed across all GEMs, including GEMs that do not contain cells, the amount of sequencing required will be vastly higher.
- ERCCs do not have 5' caps which is relevant given that the template switch reaction is mediated by the cap.
- ERCCs have a defined polyA length which does not represent the pool of endogenously expressed transcripts
- ERCCs are not associated with cellular proteins such as ribosomal proteins, which can impact the efficiency of RNA to cDNA conversion
ERCC spike-in experiments were performed with the 10x Genomics Single Cell 3’ solution (v1 chemistry) as described in Zheng et al. 2017 (https://www.nature.com/articles/ncomms14049) to assess conversion efficiencies of mRNA into cDNA. The inferred capture efficiency of the ERCC RNA spike-ins was between 6.7-8.1%, depending on the ERCC transcript (Supplementary Figure 2b). With the newer Single Cell 3’ Kit v2, the conversion rates are slightly higher, between 14-15%, and for v3 30-32%. Please also refer to the article: What fraction of mRNA transcripts are captured per cell?
If you do choose to use ERCC controls, the following considerations may be helpful:
- ERCC RNA spike-ins added to the Single Cell Master Mix will distribute evenly across all ~80,000 GEMs generated per Single Cell A Chip channel.
- Since all GEMs will contain the ERCC RNAs, physically subsampling GEMs (e.g. only using a fraction of the recovered GEMs) may be required to make the sequencing cost-effective, although you will recover fewer total cells.
- Cell Ranger provides a pre-built ERCC92 reference package for alignment of ERCC transcripts (https://support.10xgenomics.com/single-cell-gene-expression/software/downloads/latest), but does not support downstream analysis. The summary output files should be disregarded and the unfiltered gene-barcode matrix should be analyzed directly.
Example: Recommended modifications to Single Cell 3’ v2 User Guide when using ERCC RNA control:
- Dilute 1:10 ERCC RNA Spike-In Mix (Thermo Fisher PN 4456740) in nuclease-free water.
- When preparing the Single Cell Master Mix (Step 1.1), add 2.3 µL of 1:10 diluted ERCC mix in place of an equal volume of water. This is equivalent to loading approximately 133,000 ERCC molecules per GEM.
- After the completion of GEM-RT incubation, remove 1.25 μL of the emulsion (approximately 1200 GEMs) and add 125 μL Recovery Agent and 30 μL Additive A.
- Slowly remove Recovery Agent from the bottom of the tube and discard, and perform the post cDNA Amplification Reaction Cleanup (Step 2.1)
- During cDNA amplification (Step 2.2), use a total of 14 PCR cycles.
- Clean up the amplified cDNA product with 0.8X SPRIselect Reagent (Beckman Coulter PN B23318).
- During the Sample Index PCR (Step 3.5), use a total of 14 PCR cycles.
Products: Single Cell Gene Expression, Single Cell Immune Profiling.