Clustering quality assessment

Since there is no superior similarity measure, which can address all the issues, the selection of different measures is problem dependent. The optimal distance similarity method can be determined from the clustering results as well as the analysis of cluster quality assessment methods (see below). An example showing the Jif-means clustering results for a subset of gene expression data published by Bhattacharjee et al. (2001) how type of dissimilarity measure can have a great impact on the final clustering results (Fig. 5.2) and the cluster quality (Table 5.1). 

Further on, the measure RMS distance that is to be optimized is a valuable point of information in itself. It is used, for example, to compare predictions with crystal structures and invaluable for clustering similar placements. However, caution must be taken to avoid problems with symmetry in the molecules. Again, the problem of correspondence must be treated carefuUy, since, for example, a rotation of 180° of a phenyl ring should not affect the result of such a quality assessment. 

Obtaining high-quality spectra is crucial to ensure confident peptide identifications. Several processing tools have been developed to alleviate the impact of bad-quality spectra (27,28). These tools normally involve, among others, (i) spectrum quality assessment, (ii) spectrum/peak filtering, and (iii) spectrum clustering (29,30). 

These methods use a quality measure to assess cluster goodness. An internal quality measure compares sets of clusters without reliance on preexisting knowledge, such as user validation or classification models. A number of external quality measures exist and are often used to rate the quality of a cluster or the performance of a clustering method. Entropy [33] and F-measure are two examples of cluster quality measures. 

Given the same dataset, different choices of preprocessing, clustering algorithms, and distance measures could lead to varied clustering results. Therefore, the assessment of cluster validity is of utmost important. However, in practice, the cluster quality is hard to evaluate, particularly in the analysis of biological data. 

There are many reviews dealing with the impact of physicochemical properties on off-target behavior, for example, toxicity of compounds due to structural and electronic makeup [5] as well as metabolic bioactivation [51]. With these reviews in hand, it is reasonable to assess the quality of a cluster based on peer literature. However, it is much better to assess the cluster quality using real data from com-poimds with high similarity to the compounds of the cluster. 

A significant percentage of any compound library will inevitably fall into small clusters unsuitable to rigorous statistical evaluation. These must be considered separately - in our case, using diversity analysis with BCUT descriptors [39] to supplement the list derived from clustering. Throughout this process, we use visualization to assess data quality, identify potential problems such as edge effects, and check trends and patterns. 

The validity discriminant discussed in this section is the descendant of an earlier cluster validity measure used by Gunderson ( ) to assess the quality of cluster configurations obtained in an application of the Fuzzy ISODATA algorithms. It is closely related to a method suggested by Sneath ( ) for testing the distinctness, i.e. separation, of two clusters, and also borrows from the ideas of Fisher s linear discriminant theory (see chapt. 4, Duda and Hart,(2 0). The validity discriminant attempts to measure the separation between the classes of a cluster configuration usually, but not necessarily, obtained by application of the FCV algorithms. A brief description follows ... 

The advent of analytical techniques capable of providing data on a large number of analytes in a given specimen had necessitated that better techniques be employed in the assessment of data quality and for data interpretation. In 1983 and 1984, several volumes were published on the application of pattern recognition, cluster analysis, and factor analysis to analytical chemistry. These treatises provided the theoretical basis by which to analyze these environmentally related data. The coupling of multivariate approaches to environmental problems was yet to be accomplished. 

From the view point of the assessment, the quality of an HPLC separation in response to changes in different system variables, such as the stationary phase particle diameter, the column configuration, the flow rate, or mobile phase composition, or alternatively, changes in a solute variable such as the molecular size, net charge, charge anisotropy, or hydrophobic cluster distribution of a protein, can be based on evaluation of the system peak capacity (PC) in the analytical modes of HPLC separations and the system productivity (Peff) parameters in terms of bioactive mass recovered throughput per unit time at a specified purity level and operational cost structure. The system peak capacity PC depends on the relative selectivity and the bandwidth, and can be defined as... 

Fluidization quality in terms of material properties, particle characteristics, and particle group behavior thus needs to be assessed on three scales gross scale of the fluidized bed (macro scale), aggregate scale of gas bubbles, and particle clusters (meso scale), and scale of the discrete, individual particles (micro scale), as described in Chapter 4. 

In summary, to better asses the quality of V. africana seeds, we propose additional standards, such as percent of clusters, density, ashes and total alkaloids (Table III). In Ghana, the seeds are purchased according to their weight. Thus, the seed clusters and seed density ate important parameters to assess the quality of voacanga seeds. 

---