Enhanced Stereochemistry

A monomer is not always a single, fully known molecule. A synthesis yields a racemate; a separation yields two fractions nobody can yet tell apart; an assignment comes months later from an X-ray structure. This article explains how the library records what is known about a stereocentre at each of those moments, how it tells such monomers apart, and how a resolved monomer stays linked to the unknown one it came from.

What a drawing can say about a stereocentre

A wedge on a drawing says which way a substituent points. Enhanced stereochemistry, written as CXSMILES or MDL V3000, says how much that wedge is to be trusted:

Declaration	CXSMILES	Meaning	A chemist's situation
Absolute	no suffix, or `\\|a:n\\|`	The configuration is exactly as drawn	The pure, assigned enantiomer
Racemic, and group	`\\|&1:n\\|`	Both configurations are present, in a mixture	A racemate, or a diastereomeric mixture at that centre
Unknown, or group	`\\|o1:n\\|`	One of the two configurations, not known which	One fraction of a separation, before assignment

The number after the letter groups centres together. |o1:2,3| says that two centres are unknown as a pair: either both are as drawn or both are inverted, which is what a relative configuration from NMR gives. |o1:2,o2:3| says that the two are independent unknowns.

Biotoolkit Monomers accepts all three declarations, as CXSMILES in the form, the import and the REST API, and as a molfile (V2000 or V3000, with STEABS, STERAC and STEREL collections) in the form, an SD file and the REST API. The drawing shows the labels abs, or1, &1 on the centres, and the SMILES shown and exported keeps the block.

A molfile without the chiral flag

A V2000 molfile has no stereo groups. It carries one switch for the whole molecule, the chiral flag. By the MDL standard, wedges with the chiral flag set mean the configuration as drawn; wedges without it mean a racemate, all the centres in one and group. A V3000 molfile states each group explicitly, and its own chiral flag only matters when it has no stereo collection.

Drawing tools do not agree on the flag. Ketcher sets it when you draw a single enantiomer; Marvin and many exported libraries leave it off for a pure enantiomer too. The same drawing can therefore reach the library as a racemate or as an absolute structure, depending on where it was saved.

Biotoolkit Monomers follows the standard, and asks when a person is there to answer:

In the monomer form, a molfile with stereocentres, no chiral flag and no stereo collection opens a question when you click Preview: register it as a racemate (&1), or as drawn.
In an import, the option Read stereo without chiral flag as racemic, on by default, decides for every record of an SD file. Turn it off for a file from a program that does not write the flag for a single enantiomer.
Through the REST API, the parameter chiralFlagAbsent takes racemic, the default, or absolute.

A molfile with the chiral flag, a molfile with stereo collections, a SMILES and a CXSMILES are explicit, and read the same way whatever the setting.

Why five monomers can share one drawing

Take L-alanine as a residue. Drawn with a wedge on the alpha carbon, the same picture can stand for:

L-alanine, absolute;
D-alanine, absolute, the mirror wedge;
DL-alanine, the racemate, &1;
a fraction of unknown configuration, or1, from flask A;
the other fraction, or1 as well, from flask B.

$One drawing, five monomers: two absolute forms, a racemate, and two unknown fractions A and B; the stereo group is part of the uniqueness key and the isomer label tells A from B$

Two mechanisms keep them apart.

The stereo group is part of the uniqueness key

The canonical key, described in Canonical SMILES and Duplicate Detection, now ends with the enhanced-stereo block, written on the canonical atom numbering. The absolute form has no block, the racemate carries |&1:n|, the unknown carries |o1:n|: three keys, three monomers. Two centres in one group and the same two centres in two groups are different keys too.

The mirror drawing of a racemate gives the same key as the racemate, as it should: &1 on a wedge or &1 on the opposite wedge both say "both". The same holds for an unknown: or1 drawn one way or the other is the same statement.

The isomer label tells A from B

That last point has a consequence. Flasks A and B are two monomers, but their drawings are the same statement, so no structure key can separate them. The isomer label does: a short text, A and B typically, that is part of the uniqueness rule next to the structure. Two monomers may share a drawing as long as their labels differ, and an unlabelled drawing is a monomer of its own.

A label is accepted only on a structure whose key carries an or group. It has nothing to tell apart on an absolute structure, and nothing on a racemate, which is one mixture rather than one of two. The application shows the field as soon as the structure typed in the form carries an or group.

Resolution: linking the assigned monomer to its unknown

When the configuration of an unknown is finally established, the resolved form is registered as a new monomer, not as an edit of the old one. Sequences registered with the unknown stay valid, its history stays readable, and the new monomer says where it comes from: its Resolved from field names the unknown it resolves.

From synthesis to resolution: the mixture is separated into Phx1 (or1, A) and Phx2 (or1, B); after elucidation Phx3 and Phx4 are registered as absolute, each resolved from its unknown, with a compatibility check on each link and a mirror-image check across the pair

The link is a statement about chemistry, so the service checks it before writing anything: the new monomer must be one of the stereoisomers the unknown stands for.

The unknown says	The resolved monomer may say
`abs` on a centre	`abs`, same configuration. A settled centre cannot be reopened.
`or` group	`abs` on every centre of the group, all as drawn or all inverted; or the group unchanged. Never half the group, never `&`.
`&` group	Either enantiomer as `abs`; one unknown enantiomer, `or`; or the group unchanged.
nothing on a centre	nothing

Both structures must have the same skeleton once stereo is removed, with the same attachment points. At least one group must change, otherwise the new monomer is a duplicate of the unknown, not a resolution. A group may not be split in two or merged with another: |o1:2,3| states a relative configuration, and |o1:2,o2:3| would forget it. A resolution may be partial: |o1:2,o2:3| to |o1:3| settles the first centre and keeps the second open.

A refused link names the centre: atom 10 (C) is absolute in the parent and inverted here, not the same molecular graph once stereo is removed, the same stereo statement as the parent: nothing is resolved.

Labelled siblings should resolve to mirror images

Flasks A and B are two unknown enantiomers, so once both are resolved, the two results are expected to be mirror images on the or group. When the second of them settles the group the same way as the first, both registrations stand, because the chemistry may be right and the labels wrong, and a warning names the two monomers and the two labels. The REST API returns it as an X-Monomer-Warning header; the application shows it in the Stereo analysis region of each monomer.

The unknown keeps its structure

An unknown that has resolved children is the statement those children were checked against. Its structure is frozen while they exist: the name, the author or the cap groups can still change, but a structure change is refused and names the children. A corrected structure is registered as a new monomer, or the children are retired first.

A resolved monomer still points at its unknown after that unknown is retired. The link is shown with a retired mention, and the history of the unknown stays readable in Version history.

The stereo comment

A resolution usually rests on a method and a confidence: an X-ray structure on a salt, a comparison with a reference sample, a stereoselective synthesis. No structure format carries such a sentence, so the application offers a Stereo comment next to Resolved from, stored as a metadata of the monomer and editable later.

Finding the stereoisomers of a monomer

Nothing in the library says "D-alanine is the enantiomer of L-alanine": the two are two monomers with two keys, and the link between them is in the keys themselves. The REST API computes it on request. Asking for the stereoisomers of a monomer returns every registered monomer with the same skeleton, in your organisation and the public library, each with what it is to the one asked about:

Relation	Meaning, from the monomer asked about
`enantiomer`	Its mirror image: every absolute centre inverted, the groups unchanged
`racemate`	The `&` mixture it is a component of
`unknownForm`	An `or` form that stands for it, among others: the unknown it could have been resolved from
`resolvedForm`	One of the stereoisomers it stands for, narrower than it: an enantiomer of a racemate, a settled form of an unknown
`labelSibling`	The same drawing under another isomer label: the other flask
`diastereomer`	Absolute on both sides, with some centres inverted and some not
`otherStereoisomer`	The same skeleton, with stereo statements neither of which covers the other

Each entry also says whether a Resolved from link exists between the two, carries the isomer label, the stereo statement, the SMILES and whether the monomer is live or retired. An application that wants to substitute a monomer by its enantiomer, or replace a racemate by one of its registered components, has what it needs in one call.

What changes for an existing library

Monomers registered with a stereo group before this release carry a key without the block, so they cannot be told from their pure enantiomer until they are re-keyed. The re-keying is a one-off administrative step run after the upgrade, which rewrites the key of every affected monomer and archives the previous state as a version; it touches nothing else. Until it has run for your organisation, registering the enantiomer of such a monomer is refused as a duplicate.

What is not covered

Enhanced stereochemistry describes tetrahedral centres. A few statements fall outside it, or outside what the library can store today. The library refuses what it cannot store rather than recording something else, and handles the rest as follows.

Case	What the library does
Pseudo-asymmetric centres (r / s)	Handled. The r and s forms are two monomers, and a mirror drawing of either is recognised as the same compound.
Meso compounds	Handled. A meso compound is its own mirror image, so `&1` or `or1` on its centres says nothing more than the drawing: the group is read as absolute, the monomer is registered as drawn, and the service says which group it dropped. A meso compound already registered as drawn is therefore a duplicate of the same compound sent with a group. A group next to a meaningful one keeps the meaningful one. A meso compound has no enantiomer among its stereoisomers.
Enhanced stereo on double bonds (V3000 `STEBABS`, `STEBRACn`, `STEBRELn`)	Refused, in the form, the import and the REST API: an E/Z mixture or an unknown geometry would otherwise be stored as the isomer drawn. Register each isomer as its own monomer. A double bond written without geometry stays undefined.
Axial chirality: atropisomers (hindered biaryls), allenes, 2,6-disubstituted spiro[3.3]heptanes	Refused, with the atom named. The canonical key cannot tell the two forms apart, so the second would be refused as a duplicate of the first. Register the structure without the axial stereo and tell the forms apart by their name.
Mixtures that groups cannot express, such as three of the four stereoisomers of a two-centre compound	Not expressible: a group always stands for a pair. Register the components separately.

How to Resolve the Stereochemistry of a Monomer: the steps in the application for each situation.
Canonical SMILES and Duplicate Detection: what the key keeps and drops.
Versioning and Deletion: what the Version history records when a label or a link changes.
Import File Format and REST API: the ISOMER_LABEL and RESOLVED_FROM columns, the isomerLabel and resolvedFrom fields, the stereo analysis and the stereoisomers endpoints.