What this site is
Thirteen published calculators sit across nine categories: dilution and concentration, molecular biology, DNA and RNA, protein, cell culture, unit conversion, laboratory practice, experimental planning, and reference data. Alongside them are three workspaces rather than single calculations — a 96-well plate planner, a reagent inventory ledger, and a set of reference tables whose values name their sources.
Everything is free, nothing is gated behind a login, and no tool is marked as a preview or a placeholder. The site is a set of pages served to your browser; there is no backend that models you.
Why it exists
Most of this arithmetic is known. A technician who has run serial dilutions for a year does not need to be told what C₁V₁ = C₂V₂ says. What is tiring, and what goes wrong at four in the afternoon with a tube in hand, is the re-derivation: the factor that should be inverted, the unit that quietly changed, the volume that was fine on paper and impossible in a well.
BiomeBench exists to make that layer boring. It is not trying to replace judgment about an experiment; it is trying to remove the small arithmetic uncertainty that sits underneath it.
Methods on the surface, not hidden
Each calculator states the relationship it applies, the assumptions it rests on, and the limits of its own validity, on the same page as its result. There is a panel for it on every tool. The literature a formula was taken from is cited rather than implied, and units are shown rather than assumed, because a number whose unit is invisible is a number nobody can check.
That is a design position, not a courtesy. A tool that will not show its method asks you to trust it, and trust is the wrong instrument at a bench.
How the arithmetic is checked
Beyond ordinary unit tests, the project keeps two verification suites that recompute calculator outputs and plate allocations from first principles — by a deliberately different numerical route, since the same transcription error cannot catch itself — and compare the two under a stated tolerance. Those suites are barred by their own source audit from importing the code they are checking.
Say plainly what that does and does not establish. It establishes that two independent implementations of a documented model agree to a tolerance. It does not establish that the model is right for your assay, and it is not validation by anyone outside the project.
Your work stays on your machine
Inputs are processed in the browser and are not sent anywhere. Work you choose to save — a layout, an inventory entry, a captured result — is written to your browser’s own local storage under keys beginning biomebench-, and it can be deleted there. The site sets no cookies and loads no analytics or advertising script.
This is a consequence of the architecture rather than a promise: with no account and no database, there is nowhere for experimental data to accumulate.
What BiomeBench does not claim
It is not a laboratory information management system, and it does not hold an instrument, order a reagent, or become a system of record. It offers no clinical, diagnostic, or therapeutic guidance. It is not monetised through subscriptions, and it does not position itself against the large commercial suite tools that a facility may already run.
It publishes no certification, no accreditation, and no endorsement, and it does not assert a legal entity behind the site. Where the project cannot answer something from its own code, the pages here say so instead of guessing.
Where it fits in a bench day
A calculation answers one question. A result can then be captured into the experiment you are working on and reused as an input elsewhere, so a dilution already worked out does not have to be retyped into a plate layout. The planner turns that into wells, checks the layout for volume and spatial consistency, and exports a portable record. The inventory answers the following morning’s question about whether the reagent is still in stock.
Plate work today is 96-well: rows A to H, columns 1 to 12. Other plate geometries are not supported rather than supported badly.