Information-based descriptors

Jelfs, S., Erfl, P. and Selzer, P. (2007) Estimation of pKa for drug-like compounds using semiempirical and information-based descriptors. Journal of Chemical Information and Modeling, 47, 450 59. 

Nowadays, more than 4000 types of descriptors are known.17 There exist different ways to classify them. With respect to the type of molecular representation used for their calculations—chemical formula, molecular graph, or spatial positions of atoms—one speaks about ID, 2D, and 3D descriptors, respectively. Descriptors can be global (describing the molecule as a whole) and local (only selected parts are considered). One could distinguish information-based descriptors, which tend to code the information stored in molecular structures, and knowledge-based (or semiempir-ical) descriptors issued from the consideration of the mechanism of action. Most of those descriptors can be obtained with the DRAGON, CODESSA PRO, and ISIDA programs. 

S. Jelfs, P. Ertl, and P. Selzer, Estimation of pK for druglike compounds using semiempirical and information-based descriptors, J. Chem. Inf. Model. 47 (2007), pp. 450-459. 

With the growing number of proteins sequenced, there is a necessity for novel techniques to analyze protein sequences in order to determine their structure and function. The most commonly used protein sequence descriptors are based on evolutionary information and physicochemical properties. Even though these methods have proven to be efficient in most cases, in cases of transmembrane proteins, they may fall short. As the vast field of transmembrane proteins largely remains unexplored with many transmembrane proteins yet to be sequenced, it is possible to obtain new protein sequences without any known homolog. In such cases, traditional sequence analysis methods based on alignment profiles would not be sufficient. The evolutionary information-based descriptors appear inadequate, and indices based on physicochemical property can cause ambiguities. Therefore, it is of considerable interest to develop novel methods based on sequence information alone to represent protein sequences. 

A particularly good selection of physical properties may be spectra, because they are known to depend strongly on the chemical structure. In fact, different types of spectra carry different kinds of structural information, NMR spectra characterize individual carbon atoms in their molecular environment. They therefore correspond quite closely to fragment-based descriptors, as underlined by the success of approaches to predict NMR spectra by fragment codes (see Section 10.2.3). 

C, E E Hodgkin and Richards W G 1993. The Utilisation of Gaussian Functions for the Rapid nation of Molecular Similarity. Journal of Chemical Information and Computer Science 32 188-192. C and I D Kuntz 1995, Investigating the Extension of Pairwise Distance Pharmacophore sures to Triplet-based Descriptors, Journal of Computer-Aided Molecular Design 9 373-379. 

Further commonly used molecular descriptors are for example, eigenvalue-based descriptors including 3D information, such as BCUT descriptors (Burden, CAS, University of Texas eigenvalues), EVA, and WHIM descriptors (weighted holistic invariant molecular descriptors). For more information about molecular descriptors, the reader is referred to a handbook by Todeschini and Consonni and a recent review by Xue and Bajorath. ... 

Classification based on three-dimensional molecular structures requires more time-consuming computations, but results in more reliable information. These descriptors are able to identify different pharmacological profiles of compounds and thus provide the chemist with novel information that is often not recognizable from the structural formulas directly. 

Tounge, B.A., Pfahler, L.B. and Reynolds, C.H. (2002) Chemical information based scaling of molecular descriptors a universal chemical scale for library design and analysis./. Chem. Inf. Comput. Sci., 42, 879-884. 

There are several ways in which molecular descriptors can be classified. The majority of descriptors are atom-based rather than field-based. The bulk of this review is focused on a discussion of atom-based descriptors. As the name implies, atom-based descriptors are based on individual atoms, with the description extending outward to incorporate information about the atom s environment. The descriptors are typically generated by analysis of 2D or 3D connection tables, and can include ID, 2D, or 3D information about the molecule. Atom-based descriptors include individual atoms, feature counts, substructural fragments, topological indices, atomic properties, pharmacophores (see Chapters 17 and 18 of this book), and calculated physicochemical properties. 

Also commonly used descriptors are some eigenvalue-based descriptors including 3D information, such as BCUT descriptors (Burden, CAS, University of Texas eigen-... 

Similarity Searching There are two types of similarity searching procedures— also called LEVS— that are classified according to the dimensionality of their feature descriptors. 2-D methods employ structural FPs or vector-based descriptors as described in Sects. 1.2.1 and 1.2.2, while the corresponding 3-D methods involve matching pharmacophores [153, 220-223] or molecular shapes [224-226]. Since 3-D methods appear to contain more stiuctural information such as stereochemistry, which in many cases is important for activity, it is surprising that 2-D methods tend to outperform or at least perform comparably to 3-D methods. There are... 

The nbo aonbo=cs keyword requests storing the CIS-level NBOs in the shared checkpoint file where guess=read will read them as initial guess for the CAS/NBO job. Note that CAS identifies the 1st excited state as nroot=2 whereas CIS uses root=l for this state. Note also that the Gaussian open-shell CAS implementation faik to provide relevant spin density information to NBO, forcing spin-averaged NBO description of reduced accuracy. This restriction strongly detracts from the potential usefulness of CAS calculations for excited-state analysis. However, illustrative use of this method allows one to see how one can still obtain useful NBO-based descriptors of the excited state despite loss of spin information. 

---