Cluster significance analysis

The advantage of unsupervised learning methods is that any patterns that emerge from the data are dependent on the data employed. There is no intervention by the analyst, other than to choose the data in the first place, and there is no attempt by the algorithm employed to fit a pattern to the data, or seek a correlation, or produce a discriminating function (see Chapter 7). Any groupings of points which are seen on a non-linear map, a principal components plot, a dendrogram, or even a 

The probability that a cluster as tight as the active cluster would have arisen by chance involves the calculation of MSD for all the other possible clusters of three compounds. The number of clusters with an MSD value equal to or less than the active MSD is denoted by A (including the active cluster) and a probability is calculated as 

If a probability level of 0.05 or less (95 per cent certainty or better) is taken as a significance level then this cluster of actives would be regarded as fortuitous. 

Chatfield, C. and Collins, A. J. (1980). Introduction to multivariate analysis. Chapman Hall, London. 

Cormack, R. M. (1971). Journal of the Royal Statistical Society, A134, 321-67. Dizy, M., Martin-Alvarez, P. J., Cabezudo, M. D., and Polo, M. C. (1992). Journal of the Science of Food and Agriculture, 60, 47-53. 

---

Cronin MTD. The use of cluster significance analysis to identify asymmetric QSAR data sets in toxicology. An example with eye irritation data. SAR QSAR Environ Res 1996 5 167-75. 

This approach, originally proposed for binary response variables [McFarland and Cans, 1986], was extended to the quantitative biological responses y, scaled between zero and one [Rose and Wood, 1998] and then called Generalized Cluster Significance Analysis (GCSA). 

McFarland, J.W. and Gans, D.J. (1990a). Cluster Significance Analysis A New QSAR Tool for Asymmeric Data Sets. Drug InfJ., 24,705-711. 

McFarland, J.W. and Gans, DJ. (1990b). Linear Discriminant Analysis and Cluster Significance Analysis. In Quantitative Drug Design. Vol. 4 (Ramsden, C.A., ed.), Pergamon Press, Oxford (UK), pp. 667-689. 

Ordorica, M.A., Velazquez, M.L., Ordorica, J.G., Escobar, J.L. and Lehmann, P.A. (1993). A Principal Component and Cluster Significance Analysis of the Antiparasitic Potency of Praziquantel and Some Analogs. Quant.Struct.-Act.Relat., 12,246-250. 

Rose, V.S. and Wood, J. (1998). Generalized Cluster Significance Analysis and Stepwise Cluster Significance Analysis with Conditional Probabilities. Quant.Struct.-Act.Relat., 17,348-356. 

Cluster significance analysis (CSA) is a related, supervised method that can be used to determine subsets of properties that cause active compounds to cluster together. ... 

Rose, V. S., Wood, J. Generalized cluster significance analysis and stepwise cluster significance analysis with conditional probabilities. Quant. Struct.-Act. Relat. 1998,17, 348-356. 

McFarland, J.W. and Cans, D.J. (1995) Multivariate data analysis of chemical and biological data cluster significance analysis, in Chemometrics Methods in Molecular Design, Vol. 2 (ed. H. vandeWaterbeemd), VCH Publishers, New York, pp. 295-308. 

Besides the classical Discriminant Analysis (DA) and the k-Nearest Neighbor (k-NN), other classification methods widely used in QSAR/QSPR studies are SIMCA, Linear Vector Quantization (LVQ), Partial Least Squares-Discriminant Analysis (PLS-DA), Classification and Regression Trees (CART), and Cluster Significance Analysis (CSA), specifically proposed for asymmetric classification in QSAR. 

Parameter focusing is a related technique, developed by Magee [634], Different 2D plots of physicochemical properties are drawn to find out which parameter combination separates active and inactive compounds to the largest extent. Therefore, the method can be applied, in contrast to QSAR analyses, also to qualitative data. Cluster significance analysis (chapter 5.3) was developed from this approach. 

McFarland, J. W. and Gans, D. J. (1990) Linear discriminant analysis and cluster significance analysis, in Quantitative Drug Design (ed. C. A. Ramsden), Pergamon Press, Oxford. 

---