Figure 1.
Differences in small molecule binding for human to rat orthologs.
(a) Summary scatter plot. Positions along axes report the median binding affinity of all compounds against the human target (x-axis) and its ortholog in rat. Points are scaled according to the number of different compounds tested against a pair of orthologs. (b) Scatter plot of individual compound binding affinities for the human targets (x-axis) and their orthologs in rat (y-axis). (c) Scatter plot of the affinities observed when comparing results from different assays of the same compound and target (1500 human targets and 1500 rat targets). (d) Distribution of differences in binding affinity between the human target and its ortholog in rat. Positive values indicate higher affinity for the human target and vice versa. (e) Distribution of the differences observed when comparing binding affinities for the same compound and target in different assays.
Table 1.
Proposed pairs of human to rat orthologs with species specific pharmacology.
Figure 2.
(a) Distributions of pairwise sequence identity between orthologs (red) and paralogs (blue).
Distributions in the front represent all pairs retrieved from Ensembl Compara and distributions in the back pairs of targets from our analysis. (b) Distributions of differences in binding affinity between the human target and its rat ortholog (red curve) and distribution of differences in binding affinity between human paralogs (blue curve). For comparison, the distribution of inter-assay differences is outlined (black curve).
Table 2.
The most frequent ligand binding domains in the ChEMBL target dictionary.
Figure 3.
Differences in small molecule binding for human paralogs.
Sequence identity was measured on three levels: (a,d) Full sequence, (b,e) sequence of the domain containing the binding site and (c,f) residues of GPCR and kinase binding sites. Scatter plots of binding affinities measured against pairs of human paralogs are shown in the top row. Points represent median affinities of all compounds measured against a given combination of targets. Box plots in the bottom row represent the distribution of measured differences in binding affinity for pairs of targets having 0–20%, 20–40%, 40–60%, 60–80% and 80–100% identity. The level of significance of differences between bins are indicated with one (), two (
) or three asterisks (
). The dashed horizontal line indicates the mean inter-assay difference for human targets.
Figure 4.
Molecular weight and absolute differences in binding affinity.
Box plots show distributions of differences in binding affinity for small molecules grouped by equally sized molecular weight bins for paralogs. Each bin contains the same number of values and lower bin limits are shown below each box. Anova type multiple testing was carried out to assess the significance of differences between neighbouring groups and levels of significance are indicated with one (), two (
) or three asterisks (
).
Figure 5.
Species specific binding of small molecules for the HRH3.
(a) Exemplary compounds from two clusters are shown. (b) Distributions of differences observed for compounds from the two respective clusters. Pyrrolidine containing componds bind the human target with higher affinity, while most compounds with an indole scaffold bind the human and rat HRH3 with equal affinity. (c) The substitution Thr119Ala is represented by a grey threonine side chain in the human receptor. The displayed ligand is the HRH1 antagonist Doxepin (5EH) and the nearest distance to Thr119 is 2.7A.