Edit on GitHub

starplast.variation

How much a gene differs between sequenced Toxoplasma strains.

ToxoDB counts SNPs per gene across every strain it has high-throughput sequencing for, split by what the substitution does to the protein. That split is the whole value of the table: a gene with many synonymous and few nonsynonymous changes is under purifying selection, and a gene with the reverse is under diversifying selection -- most likely because the immune system is looking at it.

Why counts and not a diversity statistic

A pi or a Tajima's D would be a stronger summary and would need the alignments, which are not published per gene. Counts are what ToxoDB serves, they are comparable across genes once length is accounted for, and they answer the question the slot asks -- is this gene the same in every strain.

What the numbers look like when they are right

Nonsynonymous SNPs per kb, measured on the shipped table: SRS surface antigens 112, the ROP5 / ROP18 / GRA15 virulence loci 55, all genes 30, ribosomal proteins 2.9. That ordering is the check. The SRS family is the most polymorphic thing in the genome because it is what host immunity sees, the virulence loci are the classic strain-typing markers, and the ribosomal core is conserved. A load that does not reproduce it has joined the wrong column or the wrong genes.

COLUMNS = {'Total SNPs All Strains': 'snp_total_all_strains', 'NonSynonymous SNPs All Strains': 'snp_nonsynonymous', 'Synonymous SNPs All Strains': 'snp_synonymous', 'Non-Coding SNPs All Strains': 'snp_noncoding', 'SNPs with Stop Codons All Strains': 'snp_stop_codon'}

TABLE = 'toxodb_strain_snps.tsv'

def strain_snps( base: str, resolve=None, log=<built-in function print>) -> pandas.DataFrame:

Per-gene SNP counts across sequenced strains, indexed by resolved gene id.

Zero is a real measurement here and not a missing value: 690 genes carry no SNP in any sequenced strain, which is a statement about the gene rather than about coverage. Only genes ToxoDB did not report at all are absent.