Question: How were SNVs called in the paper "Massively parallel digital transcriptional profiling of single cells"?
Answer: To call SNVs, FreeBayes was run on the BAM output from cellranger count. A custom script was used to organize all the qualified SNVs by cell barcodes and to plot the histogram of the SNV counts per cell.
Below is a description of the methods from the publication:
SNV analysis of Jurkat and 293T scRNA-seq data
SNVs were called by running Freebayes 1.0.2 (ref. 36) on the genome BAM produced by Cell Ranger. High-quality SNVs (SNV calling Qual>=100 with at least 10 UMI counts from at least two cells; indels are ignored) that were only observed in Jurkat or 293T cells (but not both) were selected. Cells were labelled as Jurkat or 293T based on Jurkat- and 293T-specific SNV counts, where the fraction of counts from the other species is <0.2. Cells with a fraction of SNV from either species between 0.2 and 0.8 are considered multiplets. The inferred multiplet rate is 2* observed multiplet rate (to account for Jurkat:Jurkat and 293T:293T multiplets).
Alternatively, you could refer to the accompanying code:
https://github.com/10XGenomics/single-cell-3prime-snp-clustering.
Please follow the README for more details.
Note: The link to source code is provided above for interested customers, but we are not staffed to answer questions about it or provide support for modifications.