Skip to content

Experiments

An experiment is the central unit of work in MINT. It has a unique code, a type (which determines the design fields and which plugins can act on it), a status that progresses through a lifecycle, optional collaborators, and accumulated analysis results.

[Screenshot: experiment detail page with header, design panel, analyze tab, and results tab]

Anatomy

FieldDescription
CodeAuto-assigned, globally unique (EXP-001, EXP-002, …). Not editable.
TitleHuman-readable label. Editable any time.
TypeSelected at creation; determines design schema (an experiment-design plugin owns this). Not editable after creation.
Statusplannedongoingcompleted. See Lifecycle.
OwnerThe user who created it.
CollaboratorsPer-experiment role overrides on top of project membership.
Design dataJSON validated against the experiment type's schema.
Analysis resultsAccumulated outputs from analysis-plugin runs, indexed by run ID.
ArtifactsUploaded files (RAW data, plates, sequence sheets, etc.).

Create an experiment

From a project page, click New experiment on the Experiments tab.

[Screenshot: new-experiment modal with type selector and design fields]

FieldNotes
TitleRequired.
TypeRequired. Populated from installed experiment-design plugins (e.g., LCMS sequence, drug-response panel). If empty, ask your admin to install a relevant plugin.
Design dataThe form here is rendered by the type's plugin — it's whatever the plugin author defined.
Collaborators (optional)Add now or later.

Click Create. You land on the new experiment in planned status.

Lifecycle

   planned ──▶ ongoing ──▶ completed
      ▲                       │
      └──────── (admin) ──────┘
StatusMeaningWho can write
plannedDesign is being filled in. No data uploaded yet.Owner, collaborators with edit role, project members per role
ongoingData is being collected; analysis can run.Same as planned
completedFinal state. Read-only by default.Owner and admins only — and only after re-opening

Most plugins gate writes on ongoing or completed; some require completed before they'll publish results downstream. Each plugin documents its own status requirements.

[Screenshot: status pill control showing the three states]

Design plugins vs analysis plugins

Two plugin types interact with experiments differently:

Experiment-design pluginAnalysis plugin
OwnsAn experiment type, with its own database schema and CRUDNone — reads existing experiments
WritesThe experiment's design_data and any owned tablesThe experiment's analysis_results and produced artifacts
Visible atExperiment creation form, design tabExperiment Analyze tab

A given experiment has exactly one design plugin (selected by its type) but can be analyzed by many analysis plugins over time.

See Plugins for the full plugin model.

Collaborators

By default, every project member can view every experiment in the project, with edit rights determined by their project role. To override that for a specific experiment, add collaborators:

Collaborator roleEffect
ViewerCan read the experiment even if they're not a project member
EditorCan edit design data and run analyses
OwnerSame as the original owner — including delete

Collaborators are stored on the experiment itself (in collaborators), not on the project — they survive even if the user is later removed from the project. See Members & roles for the underlying RBAC.

Search and filters

The experiments list inside a project supports:

FilterNotes
Statusplanned / ongoing / completed
TypeFaceted by installed experiment types
OwnerAnyone in the project
Created inDate range
Free textSearches title, code, design fields, and notes

[Screenshot: experiments list with multiple filters applied]

Soft delete vs hard delete

Deleting an experiment marks it as removed but keeps the row, the artifacts, and any plugin-owned data for 30 days. After 30 days the row is purged. Admins can restore within the grace window. After purge, the data is gone — there is no extra backup beyond what your platform admin separately runs.

Next

Plugins — design vs analysis plugins → Marketplace — install and request plugins → Members & roles — collaborators and overrides

MINT is open source. Made by the Morscher Lab.