Computerized Management of Structure-Activity Data. I. Multivariate Analysis of Biological Data

Comparative QSAR in toxicology: examples from teratology and cancer chemotherapy of aniline mustards

During the past 30 years, thousands of quantitative structure-activity relationships (QSAR) have been published for all sorts of chemicals acting on many forms of life or parts thereof (DNA, enzymes, organelles, etc.). Very little effort has been made to show the relationship among these equations. In this report, we discuss two examples, the toxicity of phenols to rats and the effect of aniline mustards on a variety of living systems, where the electronic effects in the QSAR can be correlated to QSAR from physical organic chemistry. This enables one to make better mechanistic deductions about the biological structure-activity relationships. From this, it is concluded that radicals formed from the phenols cause birth defects.

Toward a quantitative comparative toxicology of organic compounds

Correlation equations between logP (P = octanol water partition coefficient) and the biological activity of alcohols has been derived for 101 examples on all sorts of systems, from simple proteins to whole animals. This provides an overview of the toxic nature of hydrophobic compounds which can be used as a basis for comparison of more complex chemicals. About 100 examples of the hydrophobic effects of chemicals, other than alcohols, to various living systems or their parts are presented for comparison. It is clear that hydrophobic xenobiotics are toxic to almost every form of life, including humans (or parts there of).

BACOMP--database of bioactive compounds for structure-activity relationship

BACOMP database is presented for structure-activity relationship (SAR) investigations; it was realized on a BK-1300 general purpose microcomputer using the MICRO-SETOR network database management system. Some general considerations of database design are given and the models and facilities for a development of a microcomputer-based SAR oriented database are described. The database contains the following information for the bioactive compounds: chemical structures, biological activities, trade names, reference numbers and information sources. For computer representation of chemical structures SAR oriented language is used. The database software includes: system software, data capture and data editing software, information retrieval and data processing applications. The software development is done in FORTRAN IV and MACRO assembly language. The programs are written in a completely interactive mode. The information retrieval software includes 12 functions giving an information for the database as a whole and for a single compound as well. The data processing software includes 8 functions for finding common structural fragments among compounds with similar biological activity and for estimating a structural similarity between different compounds. The functions are selected from corresponding screen MENUs. The function realization results are framed as appropriate screen formats and the receipt of hardcopies is available. The database can be used to develop predictive methods in respect of the investigated biological activity.

Formulation of de novo substituent constants in correlation analysis: inhibition of dihydrofolate reductase by 2,4-diamino-5-(3,4-dichlorophenyl)-6-substituted pyrimidines

A correlation equation based solely on de novo constants was formulated for 105 2,4-diamino-5-(3,4-dichlorophenyl)-6-substituted pyrimidines acting as inhibitors of dihydrofolate reductase. An equation with seven indicator variables gives a correlation with a correlation coefficient of 0.903 and a standard deviation of 0.229. The technique used is a modification of the Free-Wilson approach. The results indicate that correlation equations with fewer parameters than the theoretical required to account for all molecular changes may often be encountered. It is also shown that cross-product terms can be used to establish the significance of cooperative substituent effects.