SMILES to Structure: Paste a String, Get an Editable Molecule
Paste a SMILES to structure and get a clean, editable molecule, then copy back a corrected SMILES, InChI, or InChIKey. A chemist's how-to round-trip.
A SMILES string shows up in a supporting-information PDF, a vendor catalog cell, or a chatbot’s answer, and the first thing you want is to see the molecule. Is it the regioisomer you expected? Did the nitro group survive the copy? Is that ring five- or six-membered? Going from SMILES to structure is one of the most common micro-tasks in chemistry, because strings are how molecules travel between systems but humans verify by looking at the drawing. This walks through pasting a SMILES into a browser editor, getting a clean depiction you can actually edit, and copying back a corrected string or identifier — the full round-trip, not just a static picture.
What "SMILES to structure" actually does
A SMILES string encodes connectivity, bond orders, charges, isotopes, and (in isomeric SMILES) stereochemistry — but it carries no 2D coordinates. So "draw from SMILES" never means recovering the original author’s figure. It means the software perceives the molecular graph and then generates a fresh layout. Two tools can produce different-but-equally-valid depictions of the same string, and neither is "the original drawing."
Step 1 — Paste the string and let it auto-detect the format
Paste the string into the input field. A good tool detects the format from the content rather than asking you to pick. A string beginning InChI= is an InChI. A multi-line block with a counts line and an M END marker is a MOL or SDF. Anything else is treated as SMILES. You should not have to tell it which is which. If you pasted a cell out of a spreadsheet and a compound name or a stray tab rode along (CCO\taspirin), the input should trim that rather than choke on it.
Step 2 — Read the clean depiction
The string has no coordinates, so the tool has to lay the structure out. A usable depiction follows the conventions you’d accept in a figure: uniform bond lengths, 120° angles at sp² centers, heteroatom labels shown with carbons implicit, conventional ring orientation, and minimal bond crossings. A raw graph with overlapping atoms and crossed bonds reads as broken even when the connectivity is correct, which is why the layout/clean-up pass is the part that matters — it’s the difference between a viewer and an editor you’d use.
This is also the moment you confirm intent. The software supports the check; it doesn’t replace it. Look at the drawing and decide whether it’s the molecule you meant. Watch the stereocenters in particular: a flat depiction with no wedge or hash at a carbon that should be defined means the stereochemistry is undefined, not that it’s achiral.
Step 3 — Edit on the canvas
The reason to use a structure editor instead of an image generator is that the result is a live object you keep working on. Add a substituent, swap a ring, fix a bond order, drop in a benzene template — the same edits you’d make drawing from scratch. This is the whole value of the round-trip: SMILES in, edit, corrected SMILES or identifier out. A view-only renderer that hands you a flat picture fails the task the moment you spot something to change.
. in it — parent.Cl for an HCl salt, or any solvate or mixture — that period is the legitimate disconnection separator between components, not a mistake. The tool should show every component and the count, and let you keep the salt or strip to the parent on purpose. A counterion that vanishes silently is worse than an error you can see.Step 4 — Copy the corrected string or an identifier
Once the structure is right, copy it in the form your next system wants. The practical hierarchy: SMILES to communicate or re-draw the molecule, canonical SMILES for a consistent within-toolkit key, InChI for unambiguous archival, and InChIKey for database search and deduplication. One caution worth internalizing: a canonical SMILES is only canonical within one toolkit’s algorithm. RDKit, OpenEye, and ChemDraw can each emit a different canonical string for the same molecule, so never assume your canonical SMILES will byte-match another tool’s. For the trade-offs between these, see when to reach for InChI versus SMILES.
The browser-based SMILES-to-structure editor does this end to end — paste, clean layout, edit on a Ketcher canvas, then copy SMILES, canonical SMILES (RDKit), InChI, or InChIKey, or export MOL/SDF and publication-quality SVG/PNG. It runs entirely client-side, so the structure never leaves your browser.
Step 5 — Export a figure when you need one
If the structure is bound for a report or manuscript, export vector SVG or a high-DPI PNG rather than screenshotting the canvas. An ACS-style preset (uniform bond width, standard font, white background regardless of your dark-mode setting) saves you the reformatting pass. The mechanics of publication output — presets, DPI, and what each journal expects — are covered in preparing publication-quality structure figures.
Edge cases worth knowing
- The string won’t render. A malformed SMILES — an unclosed ring digit, an aromatic nitrogen written
ninstead of[nH], an over-valent atom — should produce a located error naming the problem, not a blank canvas. The most common one, the pyrrole-type nitrogen that "can’t kekulize," has a one-character fix; we cover the full catalog in reading and debugging SMILES notation. - You pasted an InChIKey. An InChIKey is a one-way hash — you can search a database with it, but you can’t reconstruct the structure from it. The right move is a PubChem lookup, not a parse attempt.
- Aromatic style surprises you. A lowercase-aromatic SMILES (
c1ccccc1) has to be kekulized for depiction. Some tools render an aromatic circle where you expected alternating double bonds. Both denote benzene; it’s a depiction-style choice, not a structural change. - Multi-line MOL block. A MOL or SDF carries its own coordinates, so the tool should preserve the embedded layout rather than re-generating one — the opposite of the SMILES case.
What this doesn’t cover
A 2D editor confirms identity and lets you edit and convert; it does not give you a 3D conformer, energy minimization, or docking pose — that’s a separate modeling step. And it can’t tell you whether the SMILES you were handed was transcribed correctly in the first place; that’s exactly why looking at the depiction is the verification step. When the drawing matches what you meant and the identifier copies clean, you’re done. Paste your next string into the SMILES-to-structure editor and check it the same way.