Why does every hit in a human or coronavirus sequence contain CG?

Those genomes are depleted of CpG dinucleotides, and base composition alone does not know that. Rerun with k-let shuffles at k = 2 or the maximal-order null, which hold dinucleotides in the null.

Which null should I use?

The maximal-order Markov model is the standard of published avoidance analyses: a site of length k is judged against the record's own (k−1)-mers. Base composition is the quickest first look; the shuffle null is exact for a chosen k-let and needs no Gaussian approximation.

Is anything uploaded?

No. Every calculation runs in the browser on the records in your project.

Analyze

Site avoidance: restriction sites a genome avoids or favors

Ask which restriction sites a set of genomes carries far less often, or far more often, than chance predicts. Every exact recognition site in the NEB catalog is counted on both strands of every record, set against a stated null (base composition, k-let-preserving shuffles, or the maximal-order Markov model), and the false discovery rate is controlled across every record and site at once. A site a lineage avoids is a footprint of the restriction systems it has lived with, and a site a design can count on being rare.

Open Site avoidance: restriction sites a genome avoids or favors Open the Site avoidance sample

Eleven real genomes, from E. coli and phage λ to SARS-CoV-2 and human chromosomes, swept for the restriction sites each one avoids

What it does

01Observed and expected counts of every exact recognition site in the NEB catalog, on both strands, in every record of a set
02Three nulls: base composition with Poisson tails, k-let-preserving shuffles with empirical tails and their Monte Carlo interval, and the maximal-order Markov model with the R'MES score
03Benjamini–Hochberg or Benjamini–Yekutieli control of the false discovery rate across every (record, site) test, with the reason the chosen one is valid
04Avoided and enriched sites ranked by how far observed departs from expected
05Every occurrence's coordinate and strand, on request

What it runs on

Checked against seqinr and R for the base-composition null, uShuffle's own C for the shuffles, and R'MES 3.1.0 for the maximal-order scores, on every site of eleven real genomes. The panel runs on @ports/workflow.

Everything runs client-side. Files are parsed in the browser and projects are saved on this machine, so an unpublished sequence is never uploaded.

Questions

Why does every hit in a human or coronavirus sequence contain CG?: Those genomes are depleted of CpG dinucleotides, and base composition alone does not know that. Rerun with k-let shuffles at k = 2 or the maximal-order null, which hold dinucleotides in the null.
Which null should I use?: The maximal-order Markov model is the standard of published avoidance analyses: a site of length k is judged against the record's own (k−1)-mers. Base composition is the quickest first look; the shuffle null is exact for a chosen k-let and needs no Gaussian approximation.
Is anything uploaded?: No. Every calculation runs in the browser on the records in your project.

Also searched as

restriction site avoidance · under-represented restriction sites · palindrome avoidance in bacterial genomes · word statistics in DNA · exceptional words · R'MES word scores · k-let preserving shuffle · uShuffle · CTAG avoidance · restriction-modification footprint · false discovery rate across a sweep