When you are focused on a molecular structure, small interface tasks can interrupt the work: finding the right command, selecting a precise group of atoms, changing a representation, or writing a script for an analysis. The challenge is often not knowing what you want to do—it is translating that intention into the right menus, selection syntax, or code.
SAMSON AI can help with those steps inside SAMSON. Open it through Interface > Assistant, or use Ctrl+0 on Windows and Linux, or Cmd+0 on Mac. Then choose the kind of help that matches the task: action, selection, or script.
/ in the Assistant input to see the available commands.Three kinds of tasks, three useful commands
/do is for actions in the current document. It is useful when you know the change you want but do not want to hunt through the interface for the relevant operation. For example, you can ask it to zoom to the binding site, remove water, erase surfaces, or select a ligand and apply a licorice model. You can also request a color change, such as making a Gaussian surface light blue.
/select is for describing what you want selected in natural language. This can be helpful when a selection depends on both molecular identity and spatial context. For instance, try /select all side chains within 5 angstrom of the ligand. SAMSON AI displays the corresponding Node Specification Language (NSL) and executes it. Seeing the NSL gives you a way to inspect how your request was translated, which can be useful when refining a selection or learning the selection language.
/script is for generating Python that uses the SAMSON Python API. It is the better fit when a task involves a calculation, a small tool, or a procedure you may want to adapt. Examples include translating atoms along the z direction, creating a distance-entry GUI with a Move button, or computing the receptor’s gyration radius. Generated scripts can also be combined with other Python modules, including tools for machine learning or trajectory analysis.
A simple workflow for a binding-site view
- Describe the target. Use
/select the binding site.or specify a more precise selection, such as side chains within a chosen distance of the ligand. - Make the selection readable. Ask
/do select the ligand and apply a licorice model.You can then adjust the view with another action, such as zooming to the binding site. - Check the result. Confirm that the intended nodes are selected and the displayed structure matches your goal. Natural-language requests can be ambiguous, so clear wording and a quick visual check matter.
This pattern separates the problem into manageable steps: define the molecular target, change how it is shown, and inspect the outcome. If a request is not interpreted as you intended, make it more specific or break it into smaller actions. For destructive or broad changes—such as removing water or erasing all surfaces—checking the document afterward is especially sensible.
When the task depends on a document
If you need help based on material outside the current structure, /learn can read a regular web page or PDF from a URL. You can then ask questions about the learned material with /refer. Learned information persists after SAMSON is closed until you use /forget, which clears all learned data. This is useful when you want to consult a paper or reference document while working, but keep in mind that SAMSON AI can misinterpret questions or provide inaccurate information. Verify critical scientific details against trusted sources.
SAMSON AI is based on GPT-5 in the Professional plan. As with any AI assistant, it may lack recent information, misunderstand an ambiguous request, or produce incorrect answers. Treat generated responses and scripts as assistance—not as a substitute for checking selections, code, or scientific conclusions.
For command details and further examples, see the SAMSON AI documentation.
Note: SAMSON and all SAMSON Extensions are free for non-commercial use. Get SAMSON.