Skip to main content

Scientific tool catalog

Use this catalog to find the Open-Science entry point and runtime requirements for a scientific method. It is a documentation index, not an application page that installs every listed program.

Software families and where they run

Software familyOpen-Science entrySetup to verifyUseful first check
Python standard library and plottingSession NotebookSelected Python interpreter; plotting dependenciesRead the real QC CSV and produce a small plot
R / base RSession Notebook; Settings → Runtimes → REnabled app-managed or detected interpreterPrint R version and reproduce sample summaries
Structure-prediction frameworksRelevant bundled Skill, usually ComputeCompatible GPU, packages, weights, input format and external MSA access if usedOne small valid sequence/complex and its output confidence
MPNN sequence-design programsProteinMPNN / LigandMPNN / SolubleMPNN SkillsRepository/checkpoints and Python dependencies; small CPU jobs are supported by these instructionsOne backbone with explicit fixed/design positions
Single-cell frameworksscGPT / scvi-tools SkillsAnnData, cell/gene labels, packages and suitable computeValidate input dimensions and required layers before training
Molecular structure renderingMolecule Connector / molecule viewerBuilt-in offline OpenChemLib routeSave and reopen aspirin.mol
File renderersFile previewSupported extension and preview-size limitsOpen the actual downloaded file
Remote batch softwareCompute Host and remote-compute SkillsHost access, scheduler and named environmentHost probe followed by a bounded job

See Skill directory for the complete 23-Skill method table. This page explains software readiness; it does not duplicate every Skill's procedure.

Inspect the selected environment

Use Settings → Runtimes to inspect available Python/R interpreters. Select the runtime actually bound to the session, then inspect packages there. An executable installed elsewhere on the computer is not automatically the active Notebook interpreter.

Use Runtimes to prepare an interpreter and Notebook to verify a calculation. If additional packages are needed, check installation and import in that environment. For a download failure, follow Network using the affected hostname and error.

Before following a model's install command

Read the installed Skill's exact requirements and the selected environment's supported setup route. Check package-name collisions: fair-esm and the Biohub esm implementation are distinct even though they share the esm namespace. A downloaded model's code and weights can also have different versions and access conditions.

For a local Notebook use the supported package-management flow in Scientific tools. For a remote host use Remote compute. A renderer displaying a PDB proves that it can view a structure; it does not prove AlphaFold or another prediction program is installed.

Implementation reference: manifest.json, SKILL.md, notebook-runtime.ts.

Use Background tasks to track supported long-running work and inspect delivered results. Scientific packages still need to be available in the selected runtime.