Computing Fragment-Based Properties

The methods shown above to compute fragment keys can be extended to compute fragment-based properties of molecules. The use of a relational table to define the fragments makes the computation suitable to using SQL to define the function. Rather than having the fragment parameters 

The simplest molecular property is molecular weight. The obvious fragments to use for this are atoms. It is a simple matter to define the SMARTS fragments for any atom. Table 8.2 shows the definition for a few common atoms. The full table for the first 103 atoms is shown in the Appendix. 

The following function is analogous to the fragment key function above. It uses a relational table to define fragments, a function to match SMILES and SMARTS (in this case count matches), and an aggregate SQL function to tally the results over all matched fragments. 

Select sum(weight count matches( 1,smarts)) From amw EOSQL Language SQL  

Another useful fragment-based function computes the polar surface area of a molecule using the method described by Ertl, Rohde, and Selzer.7 The SMARTS and partial surface areas for the fragments described by Ertl are shown in Table A.3 in the Appendix. That table is created as 

---

Thus, both fragment-based and whole-molecule-based approaches require approximations. The fragment approach assumes that a diverse set of substituents will yield diverse libraries. The whole-molecule approach is very slow and also assumes that similarity can be adequately characterized by very crude computations. This chapter focuses on the fragment-based approach, calculating properties on the substituents only, the site of attachment being replaced by a dummy atom. 

Rescaffolding, bioisosteric replacement, or fragment replacement all describe workflows where a part of a molecule is replaced by another chemical moiety by retaining the biological activity [13]. The aim of this process is the design of molecules with novel IP or different properties. This concept has a long history in medicinal chemistry, but recently this process is heavily supported by methods of computational chemistry. Based on the retrospective evaluation of COX inhibitors, we will discuss this approach in more detail. 

At the time of this writing, it must be conceded that there have been no fundamental principles-based mathematical model for Nafion that has predicted significantly new phenomena or caused property improvements in a significant way. Models that capture the essence of percolation behavior ignore chemical identity. The more ab initio methods that do embrace chemical structure are limited by the number of molecular fragments that the computer can accommodate. Other models are semiempirical in nature, which limits their predictive flexibility. Nonetheless, the diversity of these interesting approaches offers structural perspectives that can serve as guides toward further experimental inquiry. 

The simplest representation of a molecule, with respect to computing physicochemical properties, is to assume the property to be the sum of the property values of the individual constituent atoms, or groups of atoms. Extensive data bases (1,2) of atomic and group (fragment) property values have been compiled to facilitate implementation of this model. The most notable physicochemical properties employed in QSARs using an additive property model are ... 

As discussed in Chapter 2, most force fields are validated based primarily on comparisons to small molecule data and moreover most comparisons involve what might be called static properties, i.e., structural or spectral data for computed fixed conformations. There are a few noteworthy exceptions the OPLS and TraPPE force fields were, at least for molecular solvents, optimized to reproduce bulk solvent properties derived from simulations, e.g., density, boiling point, and dielectric constant. In most instances, however, one is left with the question of whether force fields optimized for small molecules or molecular fragments will perform with acceptable accuracy in large-scale simulations. 

---