Extract per-site flexible (≥95%) bacterial core genome alignments from bacterial whole-genome, reference-anchored alignments, such as from read mapping.
Output includes invariant sites in the output for detection of recombinant sites with ClonalFrameML
Compile with:
git clone https://github.com/moorembioinfo/BactCore.git
cd BactCore/
makeRun BactCore
BactCore <input.fasta> <output>And for strict cores (sites with no gaps):
BactCore --strict <input.fasta> <output.fasta>Or SNP-sites only from the strict core sites:
BactCore --strict --snps <input.fasta> <output.fasta>Maximum memory usage is approximately that of a single sequence, independent of the number of sequences in the alignment. As such with testing performed on Salmonella enterica, Mycobacterium tuberculosis, Streptococcus pyogenes and Escherichia coli maximum memory usage was never greater than 0.25Gb.
Runtime for 5,000, 10,000, 20,000 and 40,000 genomes was 8, 14, 28 and 55 minutes
| Alignment Dataset | BactCore Runtime | SNP-sites runtime | BactCore Memory Usage (GB) | SNP-sites Memory Usage (GB) |
|---|---|---|---|---|
| Escherichia coli (n = 585) | 53 | 118 | 0.24 | 3.28 |
| Streptococcus pyogenes (n = 1,816) | 51 | 99 | 0.08 | 2.53 |
| Salmonella enterica (n = 5,379) | 375; 395 | 535 | 0.20 | 8.45 |
| Mycobacterium tuberculosis (n = 30,917) | 1,940; 2,117 | 3,309 | 0.19 | 53.09 |
| Simulated (n = 5,000) | 462 | 2,148 | 0.21 | 22.53 |
| Simulated (n = 10,000) | 841 | 2,885 | 0.21 | 44.86 |
| Simulated (n = 20,000) | 1,681 | 5,468 | 0.21 | 89.58 |
| Simulated (n = 40,000) | 3,358 | 10,824 | 0.22 | 187.72 |
Runtime in seconds and maximum RAM usage (maxresidentK) in gigabytes (GB) for BactCore and SNP-sites. BactCore was run in both strict and relaxed core extraction modes. Where runtime differed by more than 1 second, both values are shown as (strict; relaxed). Memory usage for BactCore did not differ between modes. All runs included monomorphic sites in the output. The number of sequences (n) is shown per dataset.
Multi-fasta whole genome alignment derived from mapping to a reference and variant calling such as from snippy, snippy-core and snippy-clean_full_aln:
snippy-core --ref ref.fa snippyoutfiles
snippy-clean_full_aln core.full.aln > clean.full.alnUse utils/clean-alignment.py if the alignment contains sites other than {ATCGN-}
Strict core extraction is extremely sensitive to a minority of positions with gaps. Salmonella Typhimurium genomes (n= 5,379). The strict approach reduced the original 4,685,848bp alignment by 3,089,082 sites. After strict site exclusion 34.08% of sites remained (1,596,766). Conversely, with a relaxed site exclusion, 99.22% were retained (4,555,408). To avoid minority gaps biasing against strict extraction, below shows the rarefaction sampling from between 300 to 3100 randomly sampled genomes with a step of 100 and sampling repeated 50 times per sample size.
