Named-property queries
ADMET Predictor maintains an internal list of special names that are used to request fixed molecular properties. Examples include molecular weight, Hydrogen-bond donors and acceptors, counts of atoms, bonds, rings, etc. These internal names can be used in a variety of ADMET Predictor tasks, including spreadsheet searches, and are often used to define records within external query files.
The list below provides many of the names recognized automatically by ADMET Predictor.
ArRing Aromatic ring. [Group feature].
AtomCount Number of heavy (non-Hydrogen) atoms.
BondCount Number of bonds not attached to a Hydrogen atom.
CanSmiles Canonical SMILES representation of the molecule.
Charge Total formal charge on a molecule.
CompCount Number of connected components (distinct fragments).
ExtRing Ring atom having a neighbor atom external to the ring. [Group feature].
ExtArom Aromatic ring atom having a neighbor atom external to the ring. [Group feature].
ExtAliph Aliphatic ring atom having a neighbor atom external to the ring. [Group feature].
HAcc Hydrogen-bond acceptor atom.
HeteroCount Number of heteroatoms.
HeteroFraction Fraction of heavy atoms that are heteroatoms.
MinRing Smallest ring (by atom count). [Group feature].
MaxRing Largest ring (by atom count). [Group feature].
MaxRCFRingCount Maximum number of rings in any single ring-connected fragment (RCF).
MolFormula Molecular formula.
MolWeight Molecular weight.
Name Molecule name.
Neg Explicit negative charge.
Pos Explicit positive charge.
RCF Ring-connected fragment; essentially a fused ring system, but includes individual rings not fused to any other. [Group feature].
RCFCount Number of ring-connected fragments (RCFs).
Ring Aromatic or aliphatic ring. [Group feature].
RingCount Number of aromatic or aliphatic rings.
RotBond Rotatable bond.
Smiles SMILES representation of the molecule.
StereoCount Number of stereocenters. This includes double bonds with cis/trans stereochemistry.
StereoTHCount Number of atoms with a defined tetrahedral stereochemistry.
StereoDBLCount Number of bonds with a defined cis/trans stereochemistry.
WedgeCount Number of up and down wedge bonds.
Generally, these names are used in conjunction with the “NPQ” tag to form an encoded query (discussed in a above section). For instance, the query “NPQ MolWeight < 500” can be used to search a spreadsheet for all molecules having a molecular weight less than 500. Named-property queries can also be used within pharmacophore and SMARTS queries, as described in those sections. Note that when performing searches, the “NPQ” tag is optional; thus the query “MolWeight < 500 && Act > 50 && Name != “”” will be recognized as valid.
The named-property queries labeled as group features in the above list all describe more than a single atom. Note that by using recursive SMARTS syntax, a user can create a query that includes more than one atom in the definition but is not a group feature. For instance, the query “[$(c1ccccc1)]” cannot define a group feature since it is actually a single-atom query. In contrast, the similar query “c1ccccc1” is a six-atom query and therefore can define a group feature.
Users also have the option of defining new named-property queries using the Query Wizard, which is available on the Search using a query dialog generated using EDIT>>Query By Structure/Property.
This dialog allows users to enter a query by typing it directly into a window (in Encoded format) or pasting into it directly from a sketch application. You must also enter a name by which the query will be referenced in subsequent tasks such as substructure search. When the new query is loaded via this dialog, the query name is stored internally along with the default names and can be used in an identical fashion. ADMET Predictor remembers the name until the application is closed, after which time you must reload the query. Alternatively, if a user saves the open spreadsheet to a CTK file (*.ctk) after defining named queries, the queries are saved in the CTK file and will be re-loaded automatically when the file is re-opened.
The Search using a query dialog also allows users to load queries from a previously generated query file (*.cqf). Users can either create these files manually or use the Query Wizard button to create them automatically. Each record of this file must contain a special field to indicate the name of the query, as illustrated in the following example, which defines two distinct named-property queries:
NAMEQUERY ZincBinder
SLQ ONC=O
$$$$
NAMEQUERY KeyScaffold
SLQ n1ccccc1
$$$$
In the first line of each query entry, the NAMEQUERY tag indicates that a new named-property query is being defined. The subsequent name (e.g., “ZincBinder”) is the name by which the new query will be referenced, just as “MolWeight” is used to reference the molecular-weight query. The second line defines the query itself, and like all query-file records is in the form of an encoded query. Thus “SLQ” indicates a SMARTS pattern, and (for example) “ONC=O” is the pattern for a hydroxamic acid group. See the discussion of encoded queries, and the section on File Formats, for additional information on these topics.
It is important to note the significance of atom ordering in the above SMARTS patterns. Many ADMET Predictor features supporting named-property queries assume that each such query represents a single atom. This is true of queries embedded within SMARTS strings (e.g., “C<ZincBinder>C”) as well as queries used within pharmacophores (e.g., “ZincBinder.ZincBinder.3”). When the SMARTS language is used to define queries, it is the first atom of the SMARTS string that defines the “principal” atom of the query. Thus in the above two examples, the principal atom would be the single-bonded Oxygen (ZincBinder) and the Nitrogen (KeyScaffold).
Users can also load named-property queries from an SD file (*.sd). The format requirement for SD files is that query names be specified on the first line of each SD record. All other aspects of the definition process are identical to those described above for query files (*.cqf).
Note that user-defined query names must conform to fairly strict syntax requirements due to the fact that named-property queries can be embedded into queries of other types (such as SMARTS) and must not interfere with the appropriate string parsers. Most control characters, including all whitespace characters, are forbidden; the underscore character (“_”) is permitted. Any default name defined by ADMET Predictor (HAcc, Pos, etc.) can be overridden by the user if desired.