Analysis and Properties

Analysis and Properties

This page covers the shipped property suite, the molecular forms it analyzes, and how to read its results without overstating them.

Start with the Molecular Inspector

Select a structure and choose Analyze → Molecular Inspector…. Opening it without a selection shows an empty state. On macOS the menu is app-wide; on Windows use the document window's menu. The desktop inspector is a movable native window.

Read Computed on: before using a value. The inspector can switch among available molecular interpretations. A neutralized derivative and the source salt are different inputs, even when they appear in the same analysis run.

The suite uses RDKit, an open-source chemistry engine, for source composition, charge, mass, identifiers, and descriptors. It parses the source once rather than independently inferring valence or hydrogens in several engines. NMR prediction is a separate optional plugin; see NMR Prediction.

Properties available

The following are implemented method families. A method being present does not guarantee a value for every molecule.

Family	Shipped outputs and meaning
Composition	Formula, total formal charge, element counts, connected components, explicit-isotope and radical information
Mass	Average mass using engine atomic weights; monoisotopic mass using the most abundant isotopes while respecting explicit labels
Identity	Canonical SMILES, InChI, and InChIKey; these identify structures rather than measure properties
Lipophilicity and refractivity	Wildman–Crippen estimated logP and molar refractivity
Polar surface and hydrogen bonding	Ertl topological polar surface area (TPSA), hydrogen-bond donors and acceptors, and their Lipinski variants
Flexibility and composition descriptors	Rotatable bonds, heavy atoms, heteroatoms, amide bonds, and fraction of carbon atoms that are sp³
Rings and stereocentres	Ring, aromatic/aliphatic/saturated ring, heterocycle, spiro-atom, bridgehead-atom, potential stereocentre, and unspecified stereocentre counts
Surface and topology descriptors	Labute approximate surface area, Hall–Kier alpha, Kappa shape indices, Phi flexibility index, and simple/valence Chi connectivity indices
Ion positions	Adduct m/z and composition-gated hypothetical neutral-loss positions
Isotope distribution	IsoSpec theoretical isotope envelope
Thermophysical estimates	Joback normal boiling point, critical temperature, critical pressure, and critical volume
Ionization	Recognized ionizable sites, microscopic pKa transition ladders, and macroscopic values/species distributions where supported

logP describes estimated octanol/water partitioning. TPSA is a graph-based polar surface descriptor. Keep the result's conventions, including its sulfur/phosphorus treatment and ring or rotatable-bond definitions. Joback uses group contributions and declines when groups are unparameterized or its correlation is outside range; it must not sum around missing groups.

Mass, m/z, and isotope patterns

Average and monoisotopic mass answer different questions. m/z is ion mass divided by the magnitude of charge, with ion/electron bookkeeping. Available adducts include proton, sodium, potassium, ammonium, chloride, and formate cases, plus doubly protonated/deprotonated cases. Adduct methods decline for inputs already carrying net formal charge.

Hypothetical neutral losses include water, ammonia, carbon monoxide, carbon dioxide, formic acid, methanol, and hydrogen chloride. Their gate checks composition. It does not predict fragmentation, collision-energy dependence, or observed intensity.

An IsoSpec envelope is a theoretical isotope distribution, not a predicted mass spectrum. Neutral positions use daltons; drawn ions use m/z with electron correction and division by absolute charge. Intensities are normalized to a base peak of 100. Keep the truncation rule and covered-probability disclosure with the output. Explicit isotope labels are treated as certain labels, not enrichment percentages; unsupported labels decline.

IsoSpec abundance disclosure: its built-in ¹³C abundance is 0.010788 versus the CIAAW representative 0.0107: 0.82% relatively higher, raising M+1 by that fraction. The tables have no upstream provenance; values are read from the shipped binary. CIAAW is the Commission on Isotopic Abundances and Atomic Weights. Keep this disclosure beside envelope-derived numbers or images. This caveat does not apply to RDKit exact-mass arithmetic.

Interpretations and pKa

Composition, charge, and mass retain the source, including counterions. Other methods can use named largest-organic-fragment, neutralized, reference-protomer, or reference-tautomer forms. A protomer differs in proton placement/charge; a tautomer differs in proton and bond placement. The transformation ledger records the derivation rather than replacing the source drawing.

The pKa system is a frozen fast aqueous baseline, intended for aqueous, approximately room-temperature organic chemistry. Dimorphite-derived patterns locate sites; a model trained on experimental assignments supplies values, with a literature Hammett relationship contributing for narrow supported series. A site-type average is not a molecule-specific measurement.

Each microscopic value describes a particular proton-loss transition at an atom and charge state. For a basic site it is the conjugate acid's pKa. Macroscopic values describe whole-molecule titration steps derived from the microstate ladder; retain its inconsistency warning. Net-neutral and entirely uncharged fractions differ for zwitterions.

Report intervals with the site, transition, charge state, basis, and interpretation. Their calibration evidence is limited to the model's own calibration corpus; external applicability and prospective abstention work remain unfinished. Do not treat an interval as a guaranteed physical range. Metal-adjacent sites and other unsupported chemistry can be unassessed. A rewritten canonical tautomer also has an unresolved atom-correspondence limitation recorded in the audit: its apparent mapping to drawn atoms is not verified.

Status and applicability

Result status	Meaning
ok	Computed result
partial	Some output available; retain missing parts and warnings
unsupported	Outside implemented method support
not-applicable	Method does not apply
failed	Computation or runtime failure
cancelled	Cancelled before a result
timed-out	Time limit reached

Only ok and partial carry values. CLI summaries additionally use not-requested when a method was not run. Missing values are null, never zero.

Applicability is separate: in-domain, borderline, out-of-domain, or undetermined. Read its reasons and unsupported features. Mechanical success does not establish scientific applicability or experimental validation.

Agent reports

The CLI analyze and MCP analyze_structure use the same suite. A small request from the repository is:

pnpm -s chemdraft analyze --smiles 'CCO' --methods 'rdkit.composition,rdkit.monoisotopic-mass,rdkit.canonical-smiles' --format md

The command works with the same arguments on macOS and Windows. Without an output file, the readable report is inside the JSON Lines result's report field. Preserve its methods, conventions, and provenance when quoting values.

See also

• Chemistry Guarantees
• Command Line CLI
• MCP Server
• NMR Prediction