Anyone can draw a benzene ring on paper; but drawing two linked three-membered rings with reasonable bond lengths and not a single hydrogen atom missing is not something you can do casually. The daily challenge for chemists is not “whether it can be drawn,” but “whether it can be computed after it is drawn.” The cheminformatics in MuchSci’s app collection does exactly this bridging: you give it a line of SMILES text, it parses the molecule into an object, and then grows both “graph” and “numbers” from the same object at once.
A recent conversation is a ready-made example. Input C1CC1C2CC2 and ask what molecule this is, and what comes back is two SVG structural formulas plus a summary. This is not an answer assembled by retrieval, but a result produced step by step through parsing, coordinate layout, rendering, and property calculation. This article breaks down that chain.
01 One molecule, why two kinds of diagrams are needed
Start with the deliverable itself.
▲ Figure 1 | The full structure with hydrogens is for atom-by-atom checking, while the skeleton structure follows the reading conventions of the literature
The two diagrams are not duplicate deliverables; they correspond to two reading scenarios. The full structure with hydrogens marks every carbon and every hydrogen, making it hard to miss anything when checking atom counts and valence; the skeleton structure omits carbons and hydrogens according to organic chemistry conventions, so the two three-membered rings and the connecting bond are clear at a glance, and it does not look out of place in a paper or group meeting presentation. For one molecule, these are two projections of the same fact.
The trouble usually comes when you have only one of them: with only the skeleton structure, beginners cannot count the hydrogens clearly; with only the full structure, they cannot grasp the ring-system relationships. Giving both together removes the guessing step.
02 From one line of text to two SVGs, how many steps in between
Breaking down this parsing process, it roughly goes as follows.
▲ Figure 2 | The real skill shows in step 3: how the 2D coordinates are laid out determines whether the diagram is usable
What deserves a few more words here is coordinate layout. SMILES describes only connectivity and contains no coordinates; to draw the structure, a set of 2D coordinates must first be calculated for each atom so that bond lengths and angles are reasonable, rings are not distorted, and atoms do not overlap. Small molecules like bicyclopropyl are fine, but once the ring system gets large, layout quality directly determines whether the structural diagram is usable. The rendering step then uses the same set of coordinates to produce one diagram in each of the two drawing styles, so the atom positions in the two diagrams match.
The interface shows what this delivery looked like:
▲ Figure 3 | The two SVGs can be previewed and saved; a one-sentence summary and follow-up suggestions are included in the delivery record
03 Diagrams are drawn; numbers are computed
Beyond the structure, it also provides a set of numerical values.
▲ Figure 4 | Three layers—composition, topology, and energy—with all values read from the same molecule object
The one-sentence summary this time is: C1CC1C2CC2 is bicyclopropyl, C₆H₁₀, molecular weight 82.15; two isolated three-membered rings are connected by one C–C single bond, and it is not a spiro, bridged, or fused ring; all sp³, no heteroatoms, no chirality; ring strain is about 27 kcal/mol. In this passage, diagrams and numbers are inseparable—the molecular formula, ring strain, and other quantities are all calculated from the same parsed object, so there is no mismatch where “the diagram is one molecule and the numbers are another.” It also lists what can be done next: SMILES normalization and batch parsing, fingerprint and similarity calculation, multi-conformation search plus MMFF/UFF energy ranking, and 3D SDF output.
04 Hand-drawing a single molecule is fine, but batch processing is where it shines
If one molecule is drawn wrong, you can redo it; what about an entire molecule library?
▲ Figure 5 | In drug discovery, molecule libraries often contain tens of thousands of entries; hand-drawing them one by one is unrealistic
The inputs for virtual screening and similarity clustering are batches of SMILES. At this point, “looking good” is only a downstream need; the main workload is upstream parsing, normalization, deduplication, and batch descriptor calculation. This is also where cheminformatics differs from drawing software: drawing software delivers an image, while a toolchain delivers a rerunnable workflow—run the same input again, and the diagrams and numbers are consistent, which is what makes results reproducible. MuchSci brings this chain into the conversation; what it saves is the part where you set up the environment yourself, not your judgment of the results.
05 Boundaries
As usual, let’s put the boundaries on the table.
① This article discusses functional role and mode of operation; it does not constitute a commitment to the quality of results. Whether parsing succeeds and whether the coordinate layout is reasonable both depend on molecular complexity and how the input is written.
② Outputs require manual review. Structural formulas, property values, and conformational energies may all deviate due to force field or parameter choices; before use in papers or presentations, they should be checked against the original data; professional judgment cannot be outsourced.
③ For regulated areas such as drug design, the current rules of your institution and applicable regulations take precedence. This content is for research and teaching reference only.