Druglikeness is a qualitative concept used in drug design for how "druglike" a substance is with respect to factors like bioavailability.
A druglike molecule has properties such as:
A traditional method to evaluate druglikeness is to check compliance of Lipinski's Rule of Five, which covers the numbers of hydrophilic groups, molecular weight and hydrophobicity.
Since the drug is transported in aqueous media like blood and intracellular fluid, it has to be sufficiently water-soluble in the absolute sense (i.e. must have a minimum chemical solubility in order to be effective). Solubility in water can be estimated from the number of hydrogen bond donors/acceptors vs. carbon atoms in the molecule. Low water solubility translates to slow absorption and action. Too many hydrogen bond donors, on the other hand, lead to low fat solubility, so that the drug cannot penetrate the cell membrane to reach the inside of the cell.
Based on one definition, a drug-like molecule has a logarithm of partition coefficient (log P) between −0.4 and 5.6, molecular weight 160–480 g/mol, molar refractivity of 40–130, which is related to the volume and molecular weight of the molecule and has 20–70 atoms. [5]
Substructures with known toxic, mutagenic or teratogenic properties affect the usefulness of a designed molecule. However, several poisons have a good druglikeness. Natural toxins are used in pharmacological research to find out their mechanism of action, and if it could be exploited for beneficial purposes. Alkylnitro compounds tend to be irritants, and Michael acceptors, such as enones, are alkylating agents and thus potentially mutagenic and carcinogenic. [6]
Druglikeness indices are inherently limited tools. Druglikeness can be estimated for any molecule, and does not evaluate the actual specific effect that the drug achieves (biological activity). Simple rules are not always accurate and may unnecessarily limit the chemical space to search: many best-selling drugs have features that cause them to score low on various druglikeness indices. [7] Furthermore, first-pass metabolism, which is biochemically selective, can destroy the pharmacological activity of a compound despite good druglikeness.
An alternative way to estimate druglikeness is a quantitated estimate of druglikeness (QED) score, which can be calculated for molecular structures. [8] This score compresses factors in physicochemical properties calculated for a compound (including those found in Lipinski's Rule of Five), into a single, unitless score ranging from 0 (all properties unfavourable for drug candidacy) to 1 (all properties favorable), with a threshold range between the two categories in the 0.5 to 0.6 range. The advantage of this approach over rules-based methods lies in its ability to identify cases when a generally unfavourable property can be tolerated where the other parameters are close to ideal.
Druglikeness is not relevant for most biologics, since they are usually proteins that need to be injected, because proteins are digested if eaten.