neurosnap.ai tools

neurosnap.ai tools

Neurosnap.ai is RxCubed's AI & Computational Biology provider. It hosts around 145 computational tools that RxCubed can run for you — folding, docking, protein and antibody design, and a great deal else. They run on your own neurosnap.ai account, using your API key, and they spend your own credits.

This page covers everything common to all of them. For what an individual tool does, see Finding the right tool below.

Before you start

Add your key under Settings → Data & compute. Without it, Neurosnap tools appear in the Tool Selector but cannot be submitted. See API keys.

Two things to keep in mind, because they are different from RxCubed's own tools:

• They cost money. Each run spends credits from your Neurosnap account. RxCubed does not mark this up and does not pay it for you.
• Your inputs leave RxCubed. A submitted file is sent to Neurosnap and is subject to their terms and data handling. Do not submit anything you are not free to send to a third party.

Finding the right tool

Open Model → Tool Selector and search by name, or filter by provider, modality, and category. The Tool Selector reads Neurosnap's catalogue as it is now, so it is always the accurate list — newly added tools appear there, and withdrawn ones disappear.

For what a specific tool does — its method, its exact inputs, its current cost — use [Neurosnap's own tool pages](https://neurosnap.ai/). They maintain those, they are authoritative, and they are up to date. RxCubed deliberately does not keep a copy of 145 tool descriptions that would quietly drift out of date.

You can also just ask. The [AI Assistant](/docs/ai_assistant) holds an offline snapshot of the Neurosnap catalogue and can answer questions like "which Neurosnap tools do antibody design?" or "what inputs does AF2Bind need?" without a network call. Treat its answer as a starting point and confirm anything about cost or availability against Neurosnap's live page before running a large job.

Running a job

Select a tool, choose compatible project evidence for its inputs, complete the published settings, name the run, and click Run. The run is registered immediately with the provider's job ID.

File inputs come from the project's saved evidence, never from your desktop — which is what lets the run record real provenance. See Filling in a tool and collecting its results.

Checking progress and collecting results

Remote jobs take time, so they go to the Run Results tab in Tool Selector: Pending Runs while Neurosnap is still working, then Results not collected once it has finished.

A finished job's files are not saved into your project until you collect them. Nothing on that tab refreshes by itself, because every status check spends your own Neurosnap allowance — open the tab and press the check button when you are ready, then collect anything that has finished. A failed file can be retried without duplicating files that already arrived.

Where results go

Collected outputs become evidence automatically and appear in Results Hub under Category → Tool → Run → Evidence, with the run settings and selected inputs recorded as provenance. The Browse and Graph views use the same records as the tool page.

Reading the results

Predictions are hypotheses that help you decide what to test next — not evidence that something is true. This applies particularly to predicted properties and predicted structures. See Drug discovery basics for why that distinction matters, and What the computer can and cannot tell you.