Angewandte ent les general are rs(HB 10- ate hdc r potency but red by th (p) ma bons within a mo ecific lipophilic fact alone were exp ially h d.rir s of orption.solubility.cell pe eability and brain elipophilicity.and thus po d with fo and off-targe m number toler candidat ther with e ADMET cha behavior need to be optimally and appr count the entire molecular ass of th andida ave ser chemis mproving the properti of a npound as a drug can dat important ha and metrics aiming to quantitatively vide ): K=d nce for drug design are no n-pop [LLE in the drug-l able ned the S-til (QED).this tuit te ithm fref d to idat molecules anothe re )wh s the log D val lue at C is the unbound in ning the oretical and experimen al tud stability Thes and other mcdicinal cher d rule pro warrant tiny and most tructural motifs that make up thei for further me the of pi c b ular stru Matching molecular pair (MMP)analyses are large exogenous,i optimization as can point to elps sci tists to unde and and deconvolute the mecha property adj structure (e.g. alogen vs H,ester vs.OH, embark on drug discove and later durin relat hins could lead to a pow rful toolbox providin ogical targ ets and their bindin n vitr ADMET dy exist.and ent reports dem onstrate the u this ap s).the Center fo r Biologics Evalu successes Among the Protein Data Bank (PDB)and Mendeia 9130 ngewandte.orgcoefficient less than five (clogP < 5); less than five hydrogen bond donors (HBDs < 5); and less than ten hydrogen bond acceptors(HBAs <10=2 5)]. To these parameters were later added the number of rotatable bonds (NRot, averaging around six in recent years); topological polar surface area (TPSA > 75, “3–75 rule”); and flatness as measured by the fraction of sp3 carbons (Fsp3 , ratio of sp3 carbons to the total number of carbons within a molecule) (Fsp3 < 0.47).[13] These parameter limits were expected to impart on the compound favorable properties such as suitable lipophilicity for desired levels of absorption, solubility, cell permeability, and brain barrier penetration. These properties are important to and are usually correlated with formulation, delivery, and off-target selectivities linked to toxicity. Together with metabolism (e.g., CYP oxidation), the ADMET characteristics of a drug or its pharmacokinetic behavior need to be optimally and appro￾priately balanced in order for a compound to become a viable drug candidate for clinical development. For the most part, these “rules” have served medicinal chemists well in the past few years, although notable exceptions are evident. Most importantly, medicinal chemists have introduced further refinements for their drug design efforts such as “ligand efficiency” [defined as LE = 1.4 logKi /number of heavy atoms (atoms other than H);[25] where Ki = dissociation constant; relates binding energy per heavy atom to in vitro potency], “ligand-lipophilicity efficiency” [LLE,[26, 27] also known as “lipophilic efficiency” (LipE),[28] defined as LLE = LipE = log10 (Ki or IC50)log D; relates lipophilicity to in vitro potency], “ligand-efficiency-dependent lipophilic￾ity” (defined as LELP = logP/LE),[29] the “central nervous system multiparameter optimization” algorithm [referred to as CNS MPO],[30] and “lipophilic metabolism efficiency” [defined as LipMetE = logD7.4log10 (CLint,u), where log D7.4 is the log D value at pH 7.4 and CLint,u is the unbound intrinsic clearance in human liver microsomes; relates lipophilicity to metabolic stability].[31] These and other medicinal chemistry design parameters promise to provide additional tools for rational drug design as more data sets emerge and are exploited appropriately. The properties of small organic molecules are, for the most part, translations of their molecular structures, the assemblies of the various structural motifs that make up their architectures. It is, therefore, not surprising that correlations of properties with certain structural motifs have been made by analysis of available data of known drugs, compounds that failed clinical trials, preclinical drug candidates, and other ligands. Matching molecular pair (MMP) analyses are be￾coming increasingly powerful tools for lead identification and optimization purposes as they can point to significant property adjustments by small structural changes.[32–36] MMP refers to compounds differing only in relatively small features in molecular structure (e.g., halogen vs. H, ester vs. OH, Me vs. iPr). The systematic build-up of such structure–activity relationships could lead to a powerful toolbox providing correlations of structural motifs with estimates of in vitro potencies and other properties, including ADMET. Several recent reports[33–36] demonstrate the usefulness of this ap￾proach in drug discovery programs while its adoption is spreading as a consequence of its early successes. Among the most valuable general conclusions are those pertaining to lipophilicity, potency, promiscuity, and solubility. Higher lipophilicity usually leads to higher potency but also results in higher aqueous insolubility and promiscuity, both of which are liabilities for the compound. It is important to note here that lipophilic efficiency (LLE and LipE) considerations may help to understand whether potency increases are due to nonspecific lipophilic factors alone or whether specific interactions are involved. Higher numbers of aromatic, especially benzenoid, rings within the structure of a molecule increase lipophilicity, and thus potency, while at the same time lower solubility. Three aromatic rings have been suggested as the maximum number tolerable for a drug candidate, although notable exceptions exist. A better measure for this structural requirement is perhaps the Fsp3 parameter, which takes into account the entire molecular assembly of the structure. Replacement of aromatic rings with sp3 structural motifs is currently considered as a favorable feature for improving the properties of a compound as a drug candidate. Increasing numbers of chiral centers has also been recognized as a desirable feature within the structures of potential drug candidates. The rules and metrics aiming to quantitatively provide guidance for drug design are not without issues, as evidenced by recent reports questioning their absolute predictivity and validity. For example, a new measure for the “drug-likeliness” of molecules has been proposed based on desirability (desirable properties). Called the quantitative estimate of “drug-likeliness” (QED), this intuitive metric reflects the distribution of molecular properties and can be used to rank candidate molecules.[37] In another more recent report, further doubts are cast on the validity of several of the so￾called efficiency indices and metric rules for drug design.[38] Combining theoretical and experimental data, this study provides convincing analysis of a number of examples and concludes that, at the least, the majority of the proposed rules and metrics have to be viewed with skepticism, leaving LipE and the originally proposed Lipinski rules as the only guidelines warranting further scrutiny and use. The recent proliferation of such criteria and rules are indeed in want of critical evaluation and ranking themselves, pointing to the need for further improvement of the drug discovery and development process with regards to predictivity of proper￾ties based on molecular structure. Intelligence gathering on known and emerging biological targets and their ligands, whether known drugs or otherwise, small or large molecules, endogenous or exogenous, is extremely important for drug discovery. Such knowledge helps scientists to understand and deconvolute the mecha￾nism of action of both the biological targets and their ligands and provides essential information to chemists and biologists as they embark on drug discovery programs, and later during the optimization phase. A number of databases containing useful informatics on biological targets and their binding ligands already exist, and include the DrugBank database, the Therapeutic Targets Database, the U.S. FDA Orange Book (for small-molecule drugs), the Center for Biologics Evalua￾tion and Research (CBER) website (for biological drugs), the Protein Data Bank (PDB), and the Online Mendelian Angewandte . Essays 9130 www.angewandte.org 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 9128 – 9140