Frequently Asked Questions
CORINA Classic

Which are the basic concepts implemented in CORINA Classic to generate a 3D structure?

CORINA Classic is a rule- and data-based 3D molecular model builder. The system can be regarded as a fully automatic 3D model building kit. By combining mono-centric fragments with standard bond lengths and angles and by using appropriate dihedral angles a 3D model of a molecule is built. Bond lengths and angles possess only one rigid minimum and are taken from tables. Since multiple solutions exist for torsion angles, two major problems arise. First, in ring systems only restricted sets of torsion angles are allowed that ensure proper ring closure. Secondly, non-bonded interactions due to flexible chain portions have to be minimized. Therefore, CORINA Classic handles ring and chain portion in a molecule separately.

Rings of up to a size of nine atoms are processed by using a table of allowed, single ring conformations that implicitly ensure ring closure. In the case of fused or bridged systems, a backtracking search procedure finds a contradiction-free set of conformations for each single ring following some geometric and energy restrictions.

The generated ring conformations are further refined applying a simplified "pseudo" force field that contains only special geometric terms to relax the ring geometries and to identify a low-energy conformation.

For acyclic fragments and molecules, the principle of longest pathways has been implemented. The main chains are extended as much as possible by setting the torsion angles to anti or trans configurations, unless a cis double bond is specified. This method effectively minimizes non-bonding interactions, such as atom overlaps or close contacts.

After the combination of the three-dimensional fragments of the ring systems and of the acyclic parts, the entire 3D model is checked for overlapping atoms and for close contacts. If such situations are detected, CORINA Classic performs a reduced conformational analysis in order to avoid these interactions. It varies the torsion angles of strategic bonds in the molecule applying values from small molecule crystal structures until the non-bonded interactions are solved.

Finally, the 3D molecular model is output in the required file format.

An input file that contains SMILES or InChI alongside with other compound information (such as names, IDs, properties) may provide the name of the compound in a particular column. How can CORINA Classic be told from which column it should read the compound name and parse it to the respective name field in the output file?

The input option "-i nc#=N" can be used, where N is the number of the column in which the name of the SMILES or the InChI is stored. The name is then copied to the name field of the respective output format (e.g., a SYBYL MOL2 output file). The abbreviation "nc#" stands for "name column number".

An input file that contains SMILES or InChI alongside with other compound information (such as names, IDs, properties) may not provide the structure information (SMILES, InChI) in the first column. How can CORINA Classic be told from which column it should read the structure information?

By default, CORINA Classic expects the structure information (as a SMILES string or InChI) in the first column of the input file. The input option "-i sc#=N" can be used, if the structure information is provided in any other column than the first one. The number "N" is the number of the column in which the SMILES or the InChI is stored. The abbreviation "sc#" stands for "structure column number".

Is CORINA Classic parametrized for the entire periodic table or are there some limitations regarding the atom types that can be processed?

CORINA Classic is parametrized for the entire periodic table (up to 103 Lawrencium) and all molecules that can be expressed in a valid valence bond (VB theory) notation can be processed.

Does CORINA Classic calculate molecular energies (heat of formations) which are comparable to force field or quantum mechanical calculations?

CORINA Classic does not calculate "real" energies such as obtained from force field or quantum mechanical calculations. Internally, CORINA Classic uses symbolic energy values (that have been derived from FF calculations) and terms for ring systems and non-bonded interactions.
For ring systems, a full conformational analysis is performed in order to identify a low-energy conformation.
Finally, the ring systems are relaxed in a so-called "pseudo" force field to optimize the ring skeletons, but without calculating any "real" energy values.

Can CORINA Classic process structures that use pi interactions for bonding such as transition metal complexes, e.g., ferrocene?

CORINA Classic cannot generate 3D structures for pi complexes. It processes only molecules that can be expressed in a vald valence bond (VB theory) notation and with atoms with up to a maximum of six bonded neighbors (coordinatination number of 6).

Can CORINA Classic be used to calculate 3D structures of proteins, enzymes and large biological macromolecules and biopolymers?

In general, CORINA Classic does not have any limitations regarding the number of atoms or bonds of an input structure that should be converted into 3D. However, CORINA Classic has been designed to process small to medium sized organic (typically "drug-like") molecules. The larger a molecules gets the more the intra-molecular interactions gain in importance influencing the secondary structure of a molecule. CORINA Classic can model these interactions only to a limited extend and, therefore, is not able to correctly predict 3D structures of polymers and biopolymers such as proteins, enzymes or nucleic acids.

Are there any limitations in CORINA Classic regarding the number of atoms, bonds, the number or size of ring systems in a molecule or the number of molecules in a single input file (e.g., SD file)?

Basically, there are no limitations rearding the number of atoms or the number of ring atoms in a molecule. However, some file formats do not support more than 999 atoms (or bonds), such as the SD file (V2000).
There is also no limitation regarding the number of records (i.e., molecules) in an input file. CORINA Classic consectutively processes all records in the file and writes out the generated 3D models. Please note that hardware platforms and operating systems might have some limitations regarding the file size.

Does CORINA Classic use the 2D coordinates and the definitions of the chiral centers of a molecule and calculate the 3D structures with the correct stereo configurations from this information?

CORINA Classic does not need the 2D information (or structure depict) of a molecule but the connection table (CT) information of a molecule. This information is provided in standard file formats for chemical information (such as SD file or SMILES). CORINA Classic can also generate a 3D structure from an input SD file if all coordinates are set to zero (SMILES also does not provide 2D information).
Since the first commercial version of CORINA Classic, it is fully aware of stereochemistry.
CORINA Classic interprets stereo descriptors in SMILES (@ and @@ for tetrahedral centers; // and /\ for double bonds) and SD files (wedge symbols in the bond block; parity flags in the atom block). Furthermore, CORINA Classic is able to interpret input 3D structures and to determine the correct stereo chemistry from the 3D information.

How should CORINA Classic be cited correctly in any publications including papers, posters, oral presentations, books, etc?

The best way to cite CORINA Classic is to use the following three references.

(a) Sadowski, J.; Gasteiger, J.; Klebe, G. Comparison of Automatic Three-Dimensional Model Builders Using 639 X-Ray Structures. J. Chem. Inf. Comput. Sci. 1994, 34, 1000-1008 (DOI: 10.1021/ci00020a039)
(b) Schwab, C.H. Conformations and 3D pharmacophore searching. Drug Discovery Today: Technologies, Volume 7, Issue 4, Winter 2010, e245-e253 (DOI: 10.1016/j.ddtec.2010.10.003)
(c) 3D Structure Generator CORINA Classic, Molecular Networks GmbH, Nuremberg, Germany, www.mn-am.com

ROTATE Classic

Does ROTATE Classify identify generated conformations which have atom bumps or close contacts?

ROTATE Classic checks all conformations for non-bonded interactions and removes those having close contacts (or atom bumps) fully automatically.

Can ROTATE Classic cluster or group the generated conformations and output only one conformation per group?

ROTATE Classic can combine similar conformations into classes and represent each class by a single conformation (class representative). Two different and adjustable similarity criteria can be chosen. One crierion works in Cartesian space, i.e., is based on the RMS(XYZ) (root mean square) deviation of the Cartesian coordinates of all non-hydrogen atoms of the conformations. The second criterion uses the torsion angle space to calculate whether two conformations are similar or not, i.e., the comparison is based on the RMS(TA) deviation of the torsion angles along the rotated bonds of two conformerations.
For both methods, the RMS threshold that is used to define two conformations as similar, can be chosen by the user and both methods provide a balanced sampling of the conformational space.

Does ROTATE Classic use force field calculations to optimize the generated conformations?

ROTATE Classic does not use a classical force field algorithm for optimization, but applies a symbolic (or empirical) energy function. This symbolic energy function is derived from the torsion angle library (TAL) that contains the distribution of torsion angles of over 1,000 four-atomic torsion angle patterns. The distributions are stored in histogram (from 0 to 360 degree). The frequencies of the individual torsion angle values are used to derive a symbolic energy value for each torsion angle value for a specific torsion angle pattern and a gradient optimizer identifies the minimas.

Does ROTATE Classic perform a systemic and exhaustive conformational analysis or does it "prune" the conformational space or perform a biased search?

ROTATE Classic uses a knowledge base of preferred torsion angles of acyclic, rotatable bonds (or rotors). These torsion angles have been derived from a statistical analysis of the conformational preferences of open-chain portions in small molecule crystal structures xray taken from the CSD system (Cambridge Structural Database) and are stored in the so-called torsion angle library (TAL). Therefore, ROTATE Classic does not perform a "classical" systemic and exhaustive search, but exhaustively searches in this space of allowed torsion angles spanned by the TAL and the generated conformations are biased towards solid-state, crystal structures (xray geometries, experimentally determined).

How should ROTATE Classic be cited correctly in any publications including papers, posters, oral presentations, books, etc?

The best way to cite ROTATE Classic is to use the following three references.

(a) Renner, S.; Schwab, C.H.; Schneider, G.; Gasteiger, J. Impact of conformational flexibility on three-dimensional similarity searching using correlation vectors. J. Comp. Inf. Model. 2006, 46, 2324-2332 (DOI: 10.1021/ci050075s)
(b) Schwab, C.H. Conformations and 3D pharmacophore searching. Drug Discovery Today: Technologies, Volume 7, Issue 4, Winter 2010, e245-e253 (DOI: 10.1016/j.ddtec.2010.10.003)
(c) Conformer Generator ROTATE Classic, Molecular Networks GmbH, Nuremberg, Germany, www.mn-am.com

SYLVIA

How should SYLVIA be cited correctly in any publications including papers, posters, oral presentations, books, etc?

The best way to cite SYLVIA is to use the following two references.

(a) Boda, K.; Seidel, Th.; Gasteiger, J. Structure and reaction based evaluation of synthetic accessibility. J. Comput. Aided Mol. Des., 2007, 21, 311-325 (DOI: 10.1007/s10822-006-9099-2)
(b) Synthetic accessiblity estimator SYLVIA, Molecular Networks GmbH, Nuremberg, Germany, www.mn-am.com