Radar plot

At Novartis, so-called BioavailabiUty Radar Plots [44] are used to visually display the oral absorption potential of molecules. On these plots five important calculated descriptors (log P, molecular weight, PSA, number of rotatable bonds and water solubility score [45]) are displayed on the axes of a pentagonal radar plot and compared with predefined property limits (green area) which were determined by the analysis of marketed oral drugs. These plots provide an intuitive tool that displays multiple parameters as a single chart in a straightforward but informative way, providing visual feedback about the molecule s bioavailabiUty potential (Fig. 5.5). 

Ritchie, T. Rapid visualization of bioavailability potential using simple radar, plots. Current Edge Approaches to Drug Design, 2004, March, http //www. rscmodelling.org/CEAtoD D / RitchieRadar.ppt. 

Figure 8. GPCRs described using radar plots. From left to right metastatin receptor (GPR54), neuropeptide Y2 receptor and metabotropic glutamate receptor (mGluRS).

Figure 10. Radar plots of the serotonin 5HT2a receptor and ion channel domains likely to be affected by the 5HT2a antagonist sertindole (D4Navl.2 and D4Cav 1.2) and those of two dissimilar channel domains (D3Cav3.3 and DlNavl.5).

Figure 3.4 Radar plot to demonstrate the physicochemical characteristics of Lipitor. The simple plot demonstrates that two parameters fall within the optimal zone (logP and PSA), while three others slightly exceed the boundaries molecular weight, water solubility and number of rotatable bonds.

Figure 10.3 Radar plots of (left) and asp (right) values for sertindole s DRY, O, and N1 interactions with each of the seven TM regions in amine GPCRs. The plots are based on the same PCM model as in Figure 10.2. As seen, the asp values (left) do not discriminate very clearly between the different receptor regions. However, the asp values (right) reveal that distinct interaction types and TM regions are responsible for selectivity, the DRY-TM2, DRY-TM6, and DRY-TM7 interactions having the largest contributions. (Reproduced from Mol. Pharm. 2002, 67, 1465-1475 by courtesy of the American Society for Pharmacology and Experimental Therapeutics).

Figure 10.4 Radar plots of csp for haloperidol, clozapine, tiospirone, and yohimbine interactions with each of the seven TM regions in amine GPCRs.

Figure 5.2 Radar plot showing the DelvoTest (SP-NT ampoules and multi-plate formats) lowest limit of detection for 10 antimicrobial compounds in bovine milk expressed as a fraction of the current EU MRLs.

Aviation Windows, instrument panels, lighting fixture eovers, radar plotting boards, canopies... 

Figure 5.11 Radar plot with GREENSCOPE indicators for the closed-loop and open-loop simulations (case 2).

While there is yet to be developed an entirely satisfactory solution to this problem, some ideas have been applied successfully. One is the use of a radar plot. This is similar to parallel coordinates except that the axes are arranged radially. Only a limited number of CVs and MVs are practicable - perhaps a maximum of around 12, so only the more important variables are included. 

To understand the user s subjective opinion of the SCADA systan, the HCI checklist was distribnted to all control room staff, and 33 Operations Engineers responded. The following sections discnss the hndings. Radar plots were selected as the most appropriate representation for the resnlts. Each plot presents an axis illustrating how the participants rated each factor. The farther away from the center the point is, the higher the factor was rated (and thns the higher in terms of usability the system was rated). The results of the graphs are described based on the results of the respondents. It should be noted that the conclusions drawn do not necessarily represent the opinions of the authors but represent an interpretation of the combined voices of the respondents. 

Neuronal networks (Rojas 1993) simulate brain functions. In sensor science, they are used to construct non-parametric, non-linear models of the results of sensor arrays. Neuronal networks are made homogeneously of elements having the same basic structure, the so-called neurons. Often three-layer networks of the feed-forward type are built, where neurons are arranged in layers (Fig. 10.6, left). The number of input neurons in such networks corresponds to the number of received sensor signals. The numbers of hidden neurons and of output neuronSy respectively, depend on conditions. The network is trained by standard samples. In this way, the number of hidden neurons can be optimized. Neuronal networks are suited pimarily to obtain qualitative information, but less to a lesser extent for quantitative analysis. Graphical representation in the form of radar plots (Fig. 10.6, right) has proven useful. 

Figure 10.6. Scheme of three-layer neuronal feed-forward network for evaluating results of sensor arrays left). Graphical representation of signal pattern as radar plot (right)... 

Input data to the algorithm are registered with the use of an Automatic Radar Plotting Aid (ARPA) and an Automatic Identification System (AIS). These information include the course and the speed of every ship, and the distances and hearings of the target ships from the own ship. For situations in restricted waters information concerning shoals, shorelines and other static obstacles are coUected from the Electronic Chart Display and Information... 

Radar Plotting Aids ARPA anti-collision system and on the form of the process model used for the control synthesis (Bole et al. 2006, Cahill 2002, Gluver Olsen 1998). 

Figure 6.3.2 (a) Bar-plots and (b) radar-plots showing the signai obtained by a fictitious array for two fictitious vapours. Each vertex of the radar-plots corresponds to each one of the individual sensors of the e-nose. 

Figure 63A Radar plots of three different essences (peach, tutti-frutti and vanilla) obtained by using an array of 11 sensors.

Therefore, in principle a given pattern is obtained for each sample analysed. Figure 6.3.4 shows radar-plots obtained with one of the e-noses developed at the University of Buenos Aires for three different essences, where each vertex of the radar is associated to one defined sensor of the array. The shape of the radar, obtained by joining the signals of each sensor, constitutes the fingerprint of the essence for this sensor array. Note that not only are the shapes of the radar plots different but also the intensities of the signals are different (see scales). 

Only in the very simple cases where very different compounds have very different odour patterns, the products can be discriminated by visual inspection of a radar plot (see Figure 6.3.4). But, in most practical cases the radar plots display very small differences, and so mathematical methods are necessary to demonstrate whether the differences are statistically significant. Therefore, in those cases the last step of the analysis is to try to discriminate mathematically between the groups of signals obtained. 

One good way to create a summary of the capabilities of different process types is to use radar plots. Figure 3 shows each of the six process types and allows for ready comparison. 

Fig. 3 A comparison of process capabilities. Comparison of basic process capabilities using radar plots. Variables are rated from 1 to 10. Higher numbers and greater surface area are indications of higher inherent process capability.

---