How it works

How it works

Spoligotyping

The direct repeat (DR) locus of M. tuberculosis complex genomes is made of 36 bp direct repeats separated by unique spacer sequences. Strains differ by which spacers they have lost. Spoligotyping (spacer oligonucleotide typing, Kamerbeek et al. 1997) records the presence or absence of 43 reference spacers as a 43-digit binary pattern. Because it was done by hybridization for decades, large databases of patterns exist, and spoligotyping remains a common first-line genotyping method, notably for M. bovis.

In silico spoligotyping looks for the same 43 spacer sequences in whole genome sequencing data, so the results can be compared with historical spoligotyping data.

Steps

flowchart LR
    A[fastq / fasta] -->|Seal: 25-mers, 1 mismatch| B[reads per spacer]
    B -->|count ≥ min-count| C[43-digit binary pattern]
    C --> D[octal and hexadecimal codes]
    C -->|Mbovis.org database| E[SB number]

Loading

1. Spacer detection. Seal (BBTools) counts the reads (or contigs) that contain each spacer. The spacers are 25 bp long, so each spacer is searched as a single 25-mer, on both strands, allowing 1 mismatch (sequencing errors, SNPs). A read containing several spacers counts for each of them.
2. Presence or absence. A spacer is present if it is found in at least --min-count reads.
3. Codes. The binary pattern is converted to the standard octal and hexadecimal codes.
4. Name. The pattern is looked up in the Mbovis.org database of SB numbers (1,976 patterns, included with spoligotyper). When it is not there, the closest patterns (up to 3 spacers different) are reported.
5. Species and lineage. In the same pass, Seal also counts the reads on MTBC control regions and regions of difference; a second pass counts the reads carrying each allele of 62 lineage SNPs. See Species and lineage.

Minimum count

Present spacers are covered like the rest of the genome, and absent spacers are not found at all: with enough coverage, the choice of --min-count does not matter. The default is:

• 5 for reads. It ignores the odd read with a sequencing error or from contaminating DNA, and is safe from about 20× coverage.
• 1 for assemblies, where each spacer is found once at most.

Lower it for low coverage data (-m 2 or -m 3), and raise it for very deep sequencing if absent spacers get more than a few reads. spoligotyper warns when spacers called absent were seen in 1 to 4 reads: look at the SpacerCount column to decide.

Limitations

• Assemblies: the DR locus is repetitive, and short-read assemblers sometimes collapse or break it. A spacer split between two contigs is missed. When reads are available, type the reads.
• Mixed samples: a mix of strains gives the union of their spacers, i.e. a pattern that may not exist. spoligotyper flags them from the lineage SNPs and the regions of difference: see Species and lineage.
• Other mycobacteria: non-tuberculous mycobacteria have no DR locus, and give a pattern with no spacer. spoligotyper reports them as "MTBC not detected".
• SB numbers are only defined for the animal-adapted lineages. See the FAQ.