In a molecular model, the hard part is often not knowing what you want to inspect—it is translating that idea into the exact selection syntax and then applying the right operation. A request such as “show me the residues near the ligand” sounds simple, but may involve remembering a selection language, choosing a distance cutoff, and navigating menus before you can get to the science.
SAMSON AI offers a natural-language route for this kind of task. Its /select command turns a selection request into a query in Node Specification Language (NSL), displays the corresponding NSL, and executes it. Once the relevant nodes are selected, /do can carry out an action on the document. Keeping selection and action as distinct steps makes it easier to express and check what you intend.
A practical two-step workflow
- Describe the target. For example, enter
/select all side chains within 5 angstrom of the ligand.SAMSON AI translates the natural-language request into an NSL selection and runs it. - Apply an operation. Once the selection is made, ask for a specific action, such as
/do select the ligand and apply a licorice model.Other documented actions include zooming to the binding site, removing water, and changing the color of a surface.
This separation is useful when you are exploring a structure. First ask for a well-defined group of nodes; then decide what to do with it. If the request is broad or ambiguous, make it more specific before moving on. Naming the molecular feature, chain, or distance cutoff can help communicate your intent—for instance, /select all lysines in chain B.
/ in the Assistant input to see the available commands.When a selection is only the beginning
Not every modeling task is a one-off click. If you want to describe a repeatable operation, use /script to ask SAMSON AI to generate a Python script that uses the SAMSON Python API. The documented examples include translating all atoms along the z direction by 1 angstrom and computing the receptor’s gyration radius. Scripts can also be combined with other Python modules for tasks such as trajectory analysis or machine learning.
A useful way to choose between the commands is to ask what the next step should be: use /select when the main challenge is identifying nodes; use /do when you want SAMSON AI to perform an action in the document; and use /script when you need generated Python code for a task. These are different routes to solving different parts of a modeling workflow—not interchangeable ways to phrase the same request.
Keep a scientist’s check in the loop
Natural language can be ambiguous, and the documentation notes that SAMSON AI may misinterpret input or provide inaccurate information. Treat the generated selection and any resulting changes as something to review, particularly when the selection affects an important analysis or a large part of a document. Clear wording helps, but it does not remove the need to verify that the selected nodes and intended operation match your scientific question.
If you need help with concepts outside the current document, /learn can read a regular web page or PDF from a URL. You can then ask questions about learned material with /refer; /forget clears the learned data. This can be handy when you want document-aware assistance alongside a source you provide, while remembering that AI answers still need appropriate checking.
To open the assistant, use Interface > Assistant, or press Ctrl+0 on Windows and Linux or Cmd+0 on Mac. For command details and limitations, see the SAMSON AI documentation.
Note: SAMSON and all SAMSON Extensions are free for non-commercial use. Get SAMSON.