dSNAP: a computer program to cluster and classify Cambridge Structural Database searches

Barr, G.; Dong, W.; Gilmore, C.J.; Parkin, A.; Wilson, C.C.

doi:10.1107/S0021889805021308

Figure 4
The variables space corresponding to Fig. 3

viewed by selecting the `Variables' tab on the left-hand side of the screen. (a) The well chart. (b) The dendrogram with the software-selected cut level. (c) The MMDS plot. Each sphere represents an angle or distance in the FR₁C=CR₂F fragment. Spheres which are close to each other are highly correlated variables. The easiest one to see in this context is the pair coloured purple at the bottom centre of the plot which correspond to angle 18 and angle 59 in the defined cluster. (It is possible to see which variables are represented by either switching on the `Label Variables' option or placing the cursor over the sphere.) (d) The variables correlation matrix. Entries having $[\left| {\rho _{ij}^v } \right|\, \gt\, 0.8]$ are highlighted. A sample scatter plot (in this case angle 18 plotted against angle 59) obtained by selecting the relevant matrix entry is superimposed. The correlation coefficient is 0.723. The outliers can be identified by placing the cursor over the relevant data point. The colours of the points are taken from the dendrogram in (b). There is an option to display the straight line derived from linear regression.

JOURNAL OF
APPLIED
CRYSTALLOGRAPHY

ISSN: 1600-5767

Volume 38| Part 5| September 2005| Pages 833-841

doi:10.1107/S0021889805021308

Search term		doi		Advanced search
Author		volume	page