Figure 4
Optimal tubulin isotype profiles. Each row of the central matrix corresponds to a particular cancer-context i = (c, a, p) – that is, the type of cancer c (with the associated distribution q(c), shown by the horizontal bar on the left whose component regions are colored based on isotypes), how the drug is delivered a, as well as patient characteristics p, when relevant – e.g., we used a “+” symbol to indicate extra protection for a particular organ. Each column corresponds to a particular type of healthy tissue t h (with its name on the bottom, just below its associated profile q(t h )). The size of the box at position (i, t h ) – for cancer-context i = (c, a, p) and healthy tissue t h – corresponds to the value of w h = w(t h ; c, a, p), specifying the constraints on the damage to t h . (Larger squares correspond to larger w h values, which means the associated organ can tolerate more damage, and so requires less protection.) The right side shows the isotype distribution for the optimal drug, and the black bars (farthest right) show how much this optimal drug damages the associated cancer.