CMC database

The used methods are proved by key comparisons between the national metrology institutes. For chemical measurements the Comit0 Consultative pour la Quantit0 de MatiLre (CCQM) has been established. The CMC database provides a reliable service for customers all over the world to establish traceability... 

Figure 6.6. Most frequently occurring frameworks in drugs (numbers indicate percentages of occurrence in CMC database). Data are taken from Bemis and Murcko (44).

A consensus definition of a drug-like molecule has been derived from the analysis of the CMC database by defining qualifying (covering 80% of drug molecules) ranges of calculated physical properties such as molecular weight, log P, molar refractivity, and number of atoms ... 

When applied to the MDDR and CMC databases, this filter classified almost 70% of molecules as drug-Uke, while in the Advanced Chemistry Development (ACD) database only 36% of molecules were found to be drug-Uke. The advantage of this filter is that it provides a detailed structural reason for classifying molecules as drugs or non-drugs. 

FIGURE 12.3 Topological frameworks for compounds in the CMC database (numbers indicate frequency of occurrence) Drug molecules are first reduced to a combination of rings and linkers forming the framework of the molecule, then atom and bond type identities are removed. 

An analysis of side chains in drug molecules, performed by the same authors showed that there are only 1,246 different side chains among molecules in the CMC database. The average number of side chains among the molecules is 4, and the average number of heavy atoms in each side chain is 2, although 66% of all side chains have only one heavy atom. Similar analysis of the same database found that the most abundant functional groups are tertiary amines, alcohols, and carboxamides. 

Among the eight proposed indices is a general drug-like index that has been derived from the analysis of the whole CMC database and seven specific drug-like indices derived from the property distributions within the single drug classes (Table P7). 

Developed using 4836 compounds from the CMC database, DLI is defined in terms of 25 structural descriptors defining the —> chemical space and based on the idea that drug-like compounds cluster into a certain region of this chemical space [Xu and Stevenson, 2000]. 

There is another very useful analysis, and that is to compare the compounds in Beilstein to the around 6000 compounds in the CMC (Current Medicinal Compounds) database. The Bemis and Murcko analysis of the CMC database in 1996 drew attention to the fact that 94% of the compounds contained at least one ring system. The results from a comparative analysis based on ring sizes is depicted in Fig. 28.2, and it is both simple and relevant. It is derived from the size of the largest ring system present in the molecular structures of the compounds and the occurrence of this maximum ring size from rings with three or more atoms. 

The almost exact overlay for the compounds from Beilstein and those from the CMC database is quite frightening. What it tells us is that almost no medicines have been developed that fall outside the pattern of ring sizes contained in Beistein. In other words, in terms of ring sizes, the medicines that have been developed are an exact mirror of the chemistry developed in the last 200 years. 

Figure 28.2 Maximum ring size comparison (BeUstein and CMC database).

Table 4 Timing Tests for Search of 5374 Models of the CMC Database, Using the Antihistamine Pharmacophore Shown. CPU Times Are in Seconds and Were Carried Out on a VAX 6610 Computer ...

---