What the journal does, and what it refuses to do.

Review is double-blind. A reviewer is never shown who wrote the paper, and an author is never shown who reviewed it.

This is enforced by what the database returns: the function a reviewer reads a manuscript through does not return the author list, so there is no page that could show it by mistake. A submission is refused without a blinded copy, because if the only copy in the system carries the authors’ names somebody will eventually see them.

Each reviewer writes twice: comments the author will read, and comments for the editor alone. The author is shown the first in full and never the second.

An editor may not decide a manuscript they wrote, and an author may not be invited to review their own work. Both are refused by the database rather than left to anybody’s good sense.

An editor with any other conflict, such as a close collaborator, a competitor or a dispute, hands the manuscript to another editor. Authors declare their own competing interests at submission, and what they wrote is printed on the article.

Everyone listed has contributed to the work and can defend it; anyone who has contributed is listed.

The corresponding author is answerable for the submission. A change to the author list after submission is made by the editor, on the corresponding author’s written request, with every author copied.

A dispute about authorship is a matter for the Centre under A11 of its terms, and it is settled there.

You may use a language model. Say so, in the statements, naming which and for what: drafting, translation, code, analysis.

A model is not an author. It cannot take responsibility, and authorship is responsibility.

The thing this journal will not publish is text a model produced that nobody has checked. The authors answer for every sentence, however it was first written.

Every article carries a data availability statement, and what it says is printed. "Available on request" is an answer the journal accepts and prints as written, so a reader can judge it.

A reviewer may ask to see the data behind a claim. A paper whose data cannot be shown to a reviewer who asks for it is not published.

Code that produced a result is expected to be available. Where it cannot be, the article says why.

An author may withdraw a manuscript at any time before a decision, from their own desk, with a reason. A withdrawn manuscript keeps its reference and is never published.

After publication, an error that does not change the conclusions is corrected and the correction is published beside the article. An error that does change them is a retraction: the article stays readable, marked, because removing it would leave every citation of it pointing at nothing.

Every change to a published article is recorded with who made it and when.

Open access: free to read for everybody, and free to publish for members of the Centre. No submission fee, no page or colour charge and no subscription. A corresponding author who is not a member pays a single publication fee of £50 on acceptance, waived on request where it would be the obstacle.

Authors keep their copyright. Every article is published under CC BY 4.0, so anyone may read, copy and build on it provided the authors are credited.

There is no price column anywhere in this journal’s records, which is the strongest statement of this available.

An article’s reference is allocated from the Centre’s own register when the manuscript is submitted and never reused, so it identifies the work permanently whether or not a DOI is ever registered against it.

Where the Centre holds a Crossref membership a DOI is registered and printed. Where it does not, the article page says so rather than leaving a blank where a reader expects an identifier.

A complaint about a decision, about the conduct of a review, or about anything else the journal has done is made to the Editor-in-Chief, and if it concerns the Editor-in-Chief, to Centre for Immunology, Genetics, and Systems Biology itself through its contact page.

A complaint is answered in writing. Where it is upheld, what changed is recorded.