Skip to main content


Figure 6 | Journal of Cheminformatics

Figure 6

From: Are phylogenetic trees suitable for chemogenomics analyses of bioactivity data sets: the importance of shared active compounds and choosing a suitable data embedding method, as exemplified on Kinases

Figure 6

Kinases that do not show the expected, negative relationship between SAC score and bioactivity distance (kinase outliers). Kinase outlier group 1 is based on distances generated from fingerprint enrichment profiles (shown in orange) whereas kinase outlier group 2 is based on distances generated from Tanimoto comparison between bioactivity fingerprints of kinases, as performed earlier by Bamborough et al. (shown in green) [21]. Whilst the number of outliers is approximately the same for both analyses, they only have 2 kinase outliers in common, NEK6 and KPCI (shown in red). Interestingly, kinases from outlier group 1 share a much higher number of active compounds within the dataset (85.77) as compared to kinases from kinase outlier group 2 (12.03) and few shared activities have been found to be one of the limitations of analyses such as the one performed here (see main text for details). The kinases are distributed in multiple branches, but especially in the TK branch (group 1) and the AGC branch (group 2).

Back to article page