Looking at the heat maps in Fig.9, the following conclusions
can be drawn:
l The top nine most active molecules have a sulfur match and
match three of the oxygen atom in the DKPES sulfate group.
l Sulfur and sulfate oxygen atom matches alone are not sufficient
for activity. From the previous scatter plot analysis (Fig.7), we
know that the sulfate tail analog alone (Fig.8; ZINC14591952)Fig. 9Code to generate heat maps showing matches of functional groups in DKPES by the ten most active
(left) and ten least active (right) molecules tested in EOG assays. Using thematplotlib.pyplot
function that was imported aspltearlier (Fig.7), we create two subplots stored in the arrayax. Using
matplotlib.pyplot’s imshow, we plot the functional group patterns of the ten most
active molecules (X[:10], the first ten elements in the sorted data array) as a heat map in left subplot
(ax[0]). Similarly, we plot the ten least active molecules (the last ten molecules in the array,X[10:])as
a heat map in the right subplot (ax[1]). As the heat maps show, all features except sulfate-oxygens are
encoded as binary variables (0: white cell background, no match; 1: light gray, match). Sulfate-oxygens refers
to the three terminal oxygens, excluding the sulfate ester oxygen. This variable has values from 0 to 3 (up to all
three terminal oxygens being matched), where black cell backgrounds correspond to three matches, dark gray
corresponds to two matches, and light gray to one match, respectively
Inferring Activity Discriminants 319