CODESSA method

The relevance of size-related properties of hERG-blocking molecules was also detected in a 2D QSAR model developed by Coi et al. [22] after the analysis of 82 compounds through the CODESSA method. These authors developed two multiparameter models with strong predictive properties, from which, besides the involvement of hydrophobic features, the importance of linearity as opposed to globularity of the hERG blockers emerged. 

The CODESSA method, implemented in the homonymous software, is a QSAR approach based on the calculation of theoretical molecular descriptors and is a procedure for finding the best multivariate linear models based on - variable selection. 

Quantum-chemical descriptors are used in several QSAR approaches, such as, for example, theoretical linear solvation energy relationships (TLSERs), - Mezey 3D shape analysis, - GIPF approach, - CODESSA method, -> ADAPT approach. 

We have evaluated three different techniques to generate QSAR models, namely Comparative Molecular Field Analysis (CoMFA), Comprehensive Descriptors for Structural and Statistical Analysis (CODESSA), and Hologram QSAR (HQSAR). More specifically they were evaluated for their utility (predictivity, speed, accuracy, and reproducibility) to predict ER binding activity quantitatively (Tong et al., 1998 Shi et al., 2001). Common to the three QSAR methods is the... 

Luan et al. (2005PR1454) developed QSPR models to predict the pK, values of a set of 74 neutral and basic drugs via hnear and nonlinear methods. A CODESSA approach was used to derive descriptors and to build linear models RBFNN was used to generate the nonlinear models. Both models used the same descriptors selected by the heuristic method (HM) the descriptors accounted for the relative nitrogen content and polarizabUity of the compounds related to the ease of protonation of the molecules. The results were fair in view of the complexity and relatively large size of the drug molecules (R > 0.6—0.7). 

Other recently published correlative methods for predicting Tg include the group interaction modeling (GIM) approach of Porter (42), neural networks (43-45), genetic function algorithms (46), the CODESSA (acronym for Comprehensive Descriptors for Structural and Statistical Analysis ) method (47), the energy, volume, mass (EVM) approach (48,49), correlation to the results of semiempirical quantum mechanical calculations of the electronic structure of the monomer (50), and a method that combines a thermodynamic equation-of-state based on lattice fluid theory with group contributions (51). 

After descriptors have been derived, CODESSA has several advanced methods for determining correlations between descriptors and the input experimental data (draining set ). These include five types of regression analysis, a principal components analysis (PCA) treatment, four different types of multivariate analysis, and a unique heuristic method (CODESSA s default approach). These methods help the user to choose significant descriptors, determine the relationships between sets of descriptors, and evaluate the statistical significance of particular models. An intuitive and powerful graphical user interface (GUI) is used for file and information management, as well as to display the results of the correlation searches. 

CODESSA s effectiveness has been illustrated in the recent literature. Such properties as gas chromatographic retention indices, polymer glass transition temperatures, critical micelle concentrations, and other general properties of organic chemicals have been predicted with good accuracy. Forthcoming papers cover critical micelle concentrations, properties of substituted pyridines, and aqueous gas solubility. A more complete review of the philosophy behind the methods in CODESSA is also available in the literature. ... 

Selecting relevant input parameters is both important and difficult for any machine learning method. For example, in QSAR, one can compute thousands of structural descriptors with software like CODESSA or Dragon, or with various molecular field methods. Many procedures have been developed in QSAR to identify a set of structural descriptors that retain the important characteristics of the chemical compounds. " These methods can be extended to SVM models. Another source of inspiration is represented by the algorithms proposed in the machine learning literature, which can be readily applied to cheminformatics problems. We present here several literature pointers for algorithms on descriptor selection. 

In 2009, Liu et al. modelled, once again, the groups of melts previously studied by different groups, in which the influence of the anion is ignored, it being the same (bromide) in all cases. Descriptors were calculated with CODESSA, and the sole novelty of this piece of work is the use of a Projection Pursuit Regression (PPR) to derive the model, along with CODESSA built-in Heuristic Method (HM), preceded by Principal Component Analysis (PCA). The authors concluded that PPR performed better than HM... 

---