How does cellranger count auto-detect chemistry?

Question: How does cellranger count auto-detect the assay chemistry?

Answer: To auto-detect the assay chemistry (--chemistry=auto, default), cellranger count samples 100k reads (from top 1M) in the FASTQ files, and maps them to provided reference. The 3' versus 5' assay configurations are inferred based on the dominant orientation of the R2 read mapping. Assignment of 3' versus 5' requires at least 1,000 confidently mapped reads and 2x the reads in one orientation versus the other. For example, if number of antisense mapped reads are 2x greater than sense reads for the R2 read, the library is inferred to be a 5' gene expression assay.

The distinction between Single Cell 3' v1, v2, v3, and LT chemistries is made based on the fraction of barcodes overlapping the whitelist for each specific chemistry.

The chemistry settings can be nested as shown below: threeprime or fiveprime auto-detect options within the 2 possible assay configurations (3' or 5'), respectively. Within 3' or 5', there are versions of the assay and alignment options that can also be directly specified.

auto:

• threeprime for Single Cell 3':
  • SC3Pv1 for Single Cell 3' v1
  • SC3Pv2 for Single Cell 3' v2
  • SC3Pv3 for Single Cell 3' v3
  • SC3Pv3LT for Single Cell 3' v3 LT

• fiveprime for Single Cell 5':
  • SC5P-PE for Single Cell 5' paired-end (where both R1 and R2 are used for alignment)
    • This can be used if you sequenced R1 longer than 81 bases
  • SC5P-R2 for Single Cell 5' R2-only (where only R2 is used for alignment)
    
    • This assumes the recommended sequencing configuration

For more information please consult the Cell Ranger documentation.