Small Molecule Drug Design

Vartika Tomar, University of Delhi, Delhi, India

Mohit Mazumder, Jawaharlal Nehru University, Delhi, India and University of Saskatchewan, Saskatoon, Canada
Ramesh Chandra, University of Delhi, Delhi, India
Jian Yang and Meena K Sakharkar, University of Saskatchewan, Saskatoon, Canada
r 2018 Elsevier Inc. All rights reserved.

Introduction

For the balanced and proper functioning of all the life sustaining processes, nature has provided our body with all the necessary
chemical components or precursors, enzymes and neurotransmitters. Despite this, due to several factors, some exogenous and some
endogenous, some machineries or bioprocesses fail to function. The exogenous factors responsible for disrupting the
normal bodily function may vary from parasitic invasion to some chemical entities. The endogenous factors include over or
under-production of few chemicals, faulty functioning of organs or any genetic or congenital factor leading to disorders like neuro-
degenerative disorders like Alzheimer’s or Parkinson’s disease resulting from the imbalance of acetylcholine and dopamine in the
central nervous system (Moore et al., 2005). Hence, to restore the normal functioning, external aids called ‘Drugs’ or ‘Medicines’ are
required and the process of designing/discovery of drugs is called drug discovery. During this process, combinations of computational,
translational, experimental and clinical models are employed to identify new potential therapeutic entities. In early days, most of the
drugs were discovered either by the identification of the active ingredients from traditional remedies or by serendipitous discovery.
Despite having knowledge of biological systems and advanced biotechnology, drug discovery and its development is still an expensive,
time consuming, laborious, complicated and inefficient process with a high attrition rate of new therapeutic discovery. Hence,
designing a drug or a molecule with desired properties and function is an important industrial challenge.

What is a Drug?

Drugs are biological or chemical entities of synthetic or natural origin, which modulate the functions of the body without causing
any new action on the body. They can be a single compound or a mixture of different compounds. Drugs function by interacting
and modifying specific ‘targets’ in our bodies to create a molecular interaction signature that can be exploited for rapid therapeutic
repurposing and discovery. They interact and bind with the targets that are complementary to them in shape and charge and work
either by stimulating or blocking activity of their targets. For example, the analgesic effect of drug Aspirin is due to inhibition of
prostaglandin biosynthesis by acetylation of cyclo-oxygenase.

What is a Small-Molecule Drug?

According to National Cancer Institute, a small molecule drug is any organic compound or substance capable of entering cells with
an ease due to its low molecular weight (below 900 Daltons). Once it enters the cells, it can affect other molecules, such as
proteins, and may cause death of cancer cells. This is different from drugs, such as monoclonal antibodies, which are not able to
enter the cells easily because of their large molecular weight. Therefore, most of the targeted therapies are small-molecule drugs or
small molecule inhibitors.

Advantages of a Small-Molecule Drug

1. Small-molecule drugs are mostly administered orally which gives them an edge over larger drugs that require invasive pro-
cedures like injections.

The simple structures and smaller sizes of the small-molecule drugs facilitate more generic competition in comparison to
complex biologic drugs. Owing to small molecular weight (i.e., less than 900 Daltons), they have easy penetration into cell
membranes or target organs, thereby giving an intra-cellular targeting advantage over extracellularly oriented, specific biological
proteins. Since small molecule drugs can be processed into easily ingestible tablets or capsule, they can be orally administrated.
This increases the treatment regime, upregulates patient satisfaction and adherence and improves efficacy. Furthermore, these small
molecule drugs have short half-life but longer shelf lives, which is of substantial therapeutic benefit especially in cases of rapid
metabolism. Also, they require easy and less vigorous manufacturing processes as compared to complex biological drugs.

Characteristics of a Drug
A drug becomes active when it binds to its biological target, usually receptors. Receptors are protein having active sites for ligand
binding. Hence, a good ligand can be designed, if the structure of such receptors and their active sites can be identified accurately.
Before designing a drug, it is important to know what features we are looking for, in an “ideal drug candidate. The drug

Encyclopedia of Bioinformatics and Computational Biology doi:10.1016/B978-0-12-809633-8.20157-X 1

2 Small Molecule Drug Design

(i) Must be safe and effective.

(ii) Should be well absorbed orally and have good bioavailability.
(iii) Should be metabolically stable and have a long half-life.
(iv) Nontoxic with minimal or no side effects.
(v) Should have selective distribution to target tissues.

Drug Discovery

Drug discovery plays an important role to the society and for any pharmaceutical industry in ‘improving the therapeutic value and
safety of agents’ by launching newer and safe drugs in the market. Unfortunately, the process of discovery and development of
drug is a long and complicated process. It involves the identification, synthesis, characterization, screening and assays of potential
candidates for therapeutic efficacy. Drug discovery and development process includes preclinical studies on cell-based and animal
models followed by clinical trials on humans, and finally moving towards the step of obtaining regulatory approval to market the
drug. Usually it takes at least 14–16 years of research with a cost of 800 million US dollars for a compound to get developed into a
drug. After establishment of the pre-clinical data and confirmation of its action and toxicity, the compound is approved for clinical
studies which takes 1–2 years before it is released into the market. After release, several post-market surveillance and pharma-
covigilance practices are maintained to identify adverse reactions or incompatibilities when used in combination therapies
(Congreve et al., 2005). Fig. 1 depicts entire drug discovery process with their tentative timelines.

Approaches for Drug Discovery/Designing

The traditional approach involves blind screening of chemical molecules obtained either from nature or synthesized in laboratories
causing long design cycles and higher production cost. In modern drug discovery the identification of hits by screening, medicinal
chemistry and optimization of these hits for enhances selectivity, metabolic stability and efficacy/potency, affinity and oral bioa-
vailability. Enhanced electivity contributes towards reduction of potential side effects and metabolic stability increases half-life
making the drug more stable. Drug discovery process can be made cost-effective and faster by using computer-aided drug design
discovery process which involves structure-based drug design using in silico approach which plays a significant role in all stages of
drug development. Since only a potential molecule is selected, this prevents late stage clinical failures thereby reducing the cost
significantly. Table 1 lists some inhibitors developed with computational chemistry and rational drug design strategies. Better
understanding of the quantitative relationship of structure and biological activity reduces the development time of the drug to
6–8 years from 10 to 16 years. This supports the reasons to develop computer-aided molecular design (CAMD), towards automation
of molecular design (Blaney, 1990; Bugg et al., 1993). Fig. 2 outlines the steps involved in the drug development process.

Various Approaches Involved in Drug Designing

Virtual screening
Virtual screening is a computational method used for identifying lead compounds using a vast and diverse chemical com-
pound library as a reference. This computational method is an important tool for discovering lead compounds as it is faster,
more cost-efficient, and less resource intensive in comparison to experimental methods such as high-throughput screening
(Walters et al., 1998; Shoichet, 2004; Kitchen et al., 2004; Schneider and Bohm, 2002). Virtual screening process consists of two
steps: docking and scoring. AutoDock (Morris et al., 1998; Huey et al., 2007) performs ligand conformational searches for
identifying potential bound conformations, and X-Score (Wang et al., 2002) is used in re-evaluating the binding affinity of the
predicted structures. Novel lead compounds in many investigations have been successful identified using the freely available
Autodock and X-score.

Docking and scoring

Docking is a computational way of predicting the most preferred orientation of one small molecule bound to a target, resulting
into a stable complex. It consists of several steps. The first step is the application of docking algorithms that pose small molecules
within the active site of the target. Through docking we aim to predict the stable drug interactions by inspecting and modelling
drug molecular interactions between drug- and target receptor molecules. Various molecular docking tools are available, including
AutoDock, FRED, eHITS, and FTDock, etc. (Maithri et al., 2016). Scoring function is used for computing non-bonded interaction
terms between the receptor and ligand atoms. Scoring functions approximate the receptor–ligand interaction energy using mul-
tivariate regression of multiple parameters such as the number of hydrogen bonds, lipophilicity, ionic interactions, entropy
penalties, etc. Scoring functions are specifically designed to complement these docking algorithms as they evaluate the interactions
between compounds and potential targets, thereby predicting their biological activity.

High-throughput screening (HTS)

In this technique, a large number of biological modulators and effectors are screened and assayed against selected and specific
targets. The principles and methods of HTS can be exploited for screening of proteins, peptides, genomics and combinatorial
 Small Molecule Drug Design 3

Fig. 1 Different phases of drug discovery process. Adapted and modified from Lombardino, J.G., Lowe, J.A., 2004. The role of the medicinal
chemist in drug discovery—then and now. Nature Reviews Drug Discovery 3, 853–862.

chemistry libraries. HTS is a high-tech way to hasten the drug discovery process, allowing quick and efficient screening of large
compound libraries at a rate of a few thousand compounds per day or per week. (Martis et al., 2011).

Homology modelling
Computational methods can be used to predict the 3D structure of target proteins when experimental structures are not present.
Since proteins with similar sequence have similar structures, comparative modelling is used to predict target structures based on a
template with similar sequences. Homology modelling is a specific kind of comparative modelling in which the same evolutionary
origin is shared by template and target proteins. Some computer programmes and web servers which are commonly used for
automated homology modelling process are PSI-PRED and MODELER (Buchan et al., 2010; Martí-Renom et al., 2000).

Molecular dynamic simulations

Owing to the dynamic nature of biomolecules, a single static structure cannot be applied to predict putative binding sites. An
assembly of target conformations starting from a single structure are obtained by applying classic molecular dynamic (MD)
simulations. In MD methods principles of Newtonian mechanics are employed to calculate trajectory of conformations of a
protein as a function of time. The problem with classic MD method is that it gets trapped in local energy minima. To overcome this
problem, various advanced MD algorithms such as conformational folding simulations (Grubmüller, 1995), targeted–MD
(Schlitter et al., 1994), replica exchange MD (Sugita and Okamoto, 1999) and temperature accelerated MD simulations (Abrams
and Vanden-Eijnden, 2010) have been employed for traversing multiple minima energy surface of proteins.

Monte Carlo simulation

The Monte Carlo simulation is based on the concept of statistical mechanics. Monte Carlo introduces a randomness in the system
that allows hopping over the energy barriers thereby preventing the system from getting stuck in the local energy minima
4 Small Molecule Drug Design

Table 1 Selected inhibitors developed with computational chemistry and rational drug design strategies

Compound Structure of the compound Therapeutic area Function Approval

name

Captopril Hypertension congestive heart failure myocardial ACE inhibitor 1975

infarction diabetic nephropathy

Cimetidine Treatment of heartburn and peptic ulcers H2-receptor 1978

antagonist

Dorzolamide Antiglaucoma agent Antiglaucoma 1989

agent carbonic
anhydrase
inhibitor

Imatinib Chronic myeloid leukemia Tyrosine kinase 1990

inhibitor

Saquinavir Antiretroviral drug used to treat or prevent HIV/AIDS HIV-1 protease 1995
(1st generation) inhibitor

Oseltamivir Antiviral (to treat influenza A and influenza B) Influenza 1996

neuraminidase
inhibitor

(Continued)
 Small Molecule Drug Design 5

Table 1 Continued

Compound Structure of the compound Therapeutic area Function Approval

name

Indinavir Antiretroviral drug used to treat HIV/AIDS HIV protease 1996

(1st generation) inhibitor

Ritonavir Antiretroviral drug used to treat HIV/AIDS HIV protease 1996

(1st generation) inhibitor

Zanamivir Antiviral (to treat influenza A and influenza B) Neuraminidase 1999

inhibitor

Nelfinavir Antiretroviral drug used to treat HIV/AIDS HIV protease 1999

(1st generation) inhibitor

(Continued )
6 Small Molecule Drug Design

Table 1 Continued

Compound Structure of the compound Therapeutic area Function Approval

name

Lopinavir Antiretroviral drug used to treat HIV/AIDS against Peptidomimetic 2000

strains that are resistant to other protease HIV protease
inhibitors (1st generation) inhibitor

Fosamprenavir Antiretroviral prodrug used to treat HIV/AIDS HIV protease 2003

(phosphorester that is rapidly and extensively inhibitor
metabolized to amprenavir) (1st generation)

Gefitinib NSCLC EGFR kinase 2003

inhibitor

Atazanavir Antiretroviral drug used to treat HIV/AIDS HIV protease 2004

(2nd generation) inhibitor

(Continued )
 Small Molecule Drug Design 7

Table 1 Continued

Compound Structure of the compound Therapeutic area Function Approval

name

Tipranavir Antiretroviral drug used to treat HIV/AIDS (tipranavir Nonpeptidic HIV- 2005
is active against strains that are resistant to other 1protease
protease inhibitors) (2nd generation) inhibitor

Erlotinib NSCLC pancreatic cancer EGFR kinase 2005

inhibitor

Sorafenib Renal cancer Liver cancer thyroid cancer VEGFR kinase 2005
inhibitor

Darunavir Antiretroviral drug used to treat HIV/AIDS Nonpeptidic HIV-1 2006

(2nd generation) protease inhibitor

Lapatinib ERBB2-positive breast cancer EGFR/ERBB2 2007

inhibitor

Abiraterone Metastatic castration-resistant prostate cancer Androgen 2011

acetate or hormone-refractory prostate cancer synthesis
inhibitor

(Continued )
8 Small Molecule Drug Design

Table 1 Continued

Compound Structure of the compound Therapeutic area Function Approval

name

Crizotinib NSCLC ALK inhibitor 2011

100ACE, angiotensin-converting enzyme; HIV, human immunodeficiency virus; AIDS, acquired immunodeficiency syndrome; EGFR, epidermal growth factor receptor; NSCLC, non-
small cell lung cancer; VEGFR, vascular epidermal growth factor receptor; ERBB2, erb-b2 receptor tyrosine kinase 2 (also known as NEU, NGL, HER2, TKR1, CD340, HER-2, MLN 19,
HER-2/neu); ALK, anaplastic lymphoma kinase.
Adapted and modified from Prada-Graciaa, D., Huerta-Yépezb, S., Moreno-Vargasb, L.M., 2016. Application of computational methods for anticancer drug discovery, design, and
optimization. Boletin Medico Del Hospital Infantil De Mexico 73 (6), 411–423.

Fig. 2 Outline of the steps involved in drug development process. Adapted and modified from Kuhn, P., Wilson, K., Patch, M.G., Stevens, R.C.,
2002. The genesis of high-throughput structure-based drug discovery using protein crystallography. Current Opinion in Chemical Biology 6 (5),
704–710.

(Allen and Tildesley, 1989). In protein dynamics, Monte Carlo simulation run is a sequence of random steps in conformation
space inside a box where the step is decided on the basis of changes in the values of energy function during a simulation.

Metropolis criterion (MCM) simulations

Since MCM requires only energy function for evaluation and not the derivate of energy function unlike traditional MD which
drives a system towards local energy minimum, conformational space is sampled faster with MCM than molecular dynamics.
For flexible docking applications such as MCDOCK, MCM simulations have been adopted (Liu and Wang, 1999).

Rational drug design

Rational drug designing is based on the principle of logical reasoning before designing any therapeutic agents. For example, to prepare a
competitive inhibitor relative to any specific target, the predicted structure should ensure that the designed molecule exhibits endo-
genous properties. The active site should be closely examined to get idea regarding the interacting amino acid residues, so that the nature
and type of substituents and the favourable position in the molecule can be predicted, resulting in better binding.

Genetic algorithms
Genetic algorithms recombine parent conformations to child conformations which provides for molecular flexibility. The best
scoring or “fittest” combinations are kept for another round of recombination during this evolutionary simulation process. Best
possible set of solutions are evolved in the process retain favourable features from one generation to the next. Rational drug design
does not involve exhaustive searches that involve impractical large combinatorial problems. The investigation of such problems is
greatly aided by genetic algorithms; a class of algorithms mimicking some of the major characteristics of Darwinian evolution.
In order to design small organic molecule with satisfying quantitative structure-activity relationship based rules (fitness), a specific
 Small Molecule Drug Design 9

algorithm called an LEA (Ligand by Evolutionary Algorithm) has been conceived. The fitness consists of a sum of constraints that
act as range properties. The LEA takes an initial set of fragments and repeatedly improves them by means of mutation and
crossover operators which are related to those involved in the Darwinian model of evolution (Dougueta et al., 2000).

Molecular fingerprint and similarity searches

Molecular fingerprinting is a way of representing molecules which can be exploited to identify and compare structurally similar
molecules or to cluster them on the basis of structural similarity. This method is based on fewer hypotheses and is less com-
putationally taxing than QSAR or pharmacophore mapping models and is more quantitative in nature since it relies entirely on
chemical structure and omits compounds with known biological activity. Moreover, fingerprint-based methods equally consider
all parts of the molecule instead of focusing on only those parts of the molecule which are thought to be most important for
activity. Hence, this method is less prone to error overfitting and requires smaller datasets to begin with. Fingerprint methods can
also be used for searching databases for compounds that are similar in structure to a lead query, thereby providing an extended
collection of compounds that can be tested for improved activity over the lead.

De Novo ligand designing

Constructing novel molecules from scratch is called de novo drug design. This approach is particularly challenging because the
compound space to be searched can often become enormous. Defining a binding site for molecules in a target and determining
what atoms or functional groups should be placed at certain loci at the binding site is one of the most critical and difficult steps of
de novo drug design. (Klebe, 2000; Bohm and Stahl, 2000).

Case Studies
Discovery of tyrosine kinase inhibitors by docking into a molecular dynamics generated inactive kinase conformation
Type II inhibitors, which are the inactive DFG-out conformation of tyrosine kinases show good binding and selective profiles with
several small molecules over other kinase targets. An explicit solvent molecular dynamics (MD) simulation of the complex of the
catalytic domain of a tyrosine kinase receptor, ephrin type-A receptor 3 (EphA3), and manual docking of type II inhibitors was
performed to obtain a set of DFG-out structures by Zhao et al. (2012). A single snapshot from the MD trajectory(for virtual
screening) was selected using the automatic docking of four previously reported type II inhibitors (Fig. 3). Since a glycine-rich loop
(Gloop), can adopt various conformations, it tightly encompasses the small type I head group of compound 1 and collapses into
the ATP binding site. As a result, the bigger type I head groups of compounds 2–4 clashes with the G-loop in the ATP binding site.
In addition, the entry of the piperazine group of compound 4 is blocked by the side chain of Tyr742 (Fig. 4). High-throughput
docking of a pharmacophore-tailored library of 175,000 molecules results in about 4 million poses, which can be further filtered
and sorted by van der Waals efficiency and force-field-based energy function. Remarkably, about 20% of the compounds with
predicted binding energy smaller than
10 kcalmol
1 are known to be type II inhibitors and besides, a series of 5-(piperazine-1-
yl)isoquinoline derivatives was identified as a novel class of low-micromolar inhibitors of both EphA3 and dephosphorylated
Abelson tyrosine kinase (Abl1). A similar affinity to the gatekeeper mutant T315I of Abl1 is suggested by the in silico binding
mode of the new inhibitors and competition binding assay studies. Additional evidence for the type II binding mode was obtained
by two 300 ns MD simulations of the complex between N-(3-chloro-4-(difluoromethoxy)-phenyl)-2-(4-(8-nitroisoquinolin-5-yl)
piperazin-1-yl)acetamide and EphA3 provided additional evidence of the type II binding mode.
Most type II inhibitors can be mapped well into three major hydrophobic interactions: ATP front site, ATP back site, and the
allosteric site (Fig. 5).

Computer-Aided Drug Design (CADD)

CADD helps scientists in minimizing the synthetic and biological testing efforts by focussing only on the most promising
compounds. Besides explaining the molecular basis of therapeutic activity, it also predicts possible derivatives that would improve
activity. CADDD entails (Kapetanovic, 2008):

(1) Drug discovery and development processes being streamlined by the use of computing power.
(2) Identification and optimization of new drugs using leverage of chemical and biological information about targets and/or ligands.
(3) In silico designing of filters for the elimination of undesirable compounds with properties like poor activity and/or poor
absorption, distribution, metabolism, excretion and toxicity, ADMET which facilitate selection of the most promising candidates.

Advantages of CADD
The main advantages of drug discovery through CADD are:

(i) For experimental testing, smaller set of compounds are selected from large compound libraries.
10 Small Molecule Drug Design

Fig. 3 Known type II inhibitors of EphA3. The values next to the compound number are the experimentally measured dissociation constant
against phosphorylated EphA3 and the predicted binding free energy. These values were calculated by using the MD-IF structure and a scoring
function with continuum solvation and hydrogen bonding penalty.

(ii) Drug metabolism and pharmacokinetics (DMPK) properties like absorption, distribution, metabolism, excretion and the
potential for toxicity (ADMET) are increased by optimization of lead compounds.
(iii) Designing of novel compounds can be achieved either by “growing” starting molecules one functional group at a time or
by piecing together fragments into novel chemotypes (Veselovsky and Ivanov, 2003).
(iv) Traditional experimentation which requires animal and human models can be replaced by CADD, saving both time and
cost (Mallipeddi et al., 2014).
 Small Molecule Drug Design 11

Fig. 4 Comparison of A) the crystal structure (PDB: 3DZQ) of the complex of EphA3 with inhibitor 1 and B) the binding mode obtained by
docking compound 4 into the MD-IF structure. For the differences in orientations of Tyr 742 and Phe 765, and in the G-loop. Adapted and
modified from Zhao, H., Huang, D., Caflisch, A., 2012. Discovery of tyrosine kinase inhibitors by docking into an inactive kinase conformation
generated by molecular dynamics. ChemMedChem 7 (11), 1983–1990.

Fig. 5 Key interactions pharmacophore mapping of type II kinase inhibitors illustrated by compound 2 and EphA3. Dashed lines are hydrogen
bonds. Pharmacophore elements used to filter the ZINC library: two acceptors (dashed circles), one donor (solid circle), and three hydrophobic
rings (ovals). Because of the flexibility and solvent exposure of the Glu 670 side chain, the hydrogen bond to Glu 670 was not used as a
pharmacophore.

(v) Reduces the chances of drug resistance and thus would lead to production of lead compounds which would target the
causative factor.
(vi) CADD also leads to the construction of high quality datasets and libraries that can be optimized for high molecular
diversity or similarity (Ou-Yang et al., 2012).

Types of CADD
The choice of CADD approaches to be employed is determined by the availability of the experimentally determined 3D structures
of target proteins. Structure-based CADD uses our knowledge of the target protein structure to calculate interaction energies,
12 Small Molecule Drug Design

whereas in ligand-based CADD, chemical similarity searches or construction of predictive, quantitative structure-activity rela-
tionship (QSAR) models exploits our knowledge of known active and inactive molecules.(Kalyaanamoorthy and Chen, 2011).
Structure based CADD combines information from several fields, for example, X-ray crystallography and/or NMR, synthetic
organic chemistry, molecular modelling, QSAR, and biological evaluation (Marrone et al., 1997). Through structure based CADD,
we aim to design compounds with strong binding affinity with the target, thereby exhibiting properties like reduced free energy,
improved DMPK/ADMET properties and target specification i.e., reduced off-target (Jorgensen et al., 2010). Virtual high-
throughput screening (vHTS) also known as screening of virtual compound libraries is one of the most common applications of
CADD Kalyaanamoorthy and Chen, 2011). Fig. 6 represents an overview of CADD drug designing/design pipeline.

Structure–based drug discovery

This method exploits knowledge of the three-dimensional structure of a receptor complexed with a lead molecule for optimization of
the bound ligand or a series of congeneric molecules. It requires the understanding of receptor–ligand interactions. The structural
information can be obtained either from X-ray crystallography, NMR, or from homology modelling. A medicinal chemist can use a
model with a given structure for computing the activity of a molecule (Lewis, 2005). Some of these approaches provide accurate binding
modes, while cater to fast searching of large databases. Some approaches of structure-based drug designing are explained below.

Structure-based virtual high-throughput screening

Structure-based virtual high-throughput screening (SB-vHTS) is an in-silico method which helps identify putative hits out of hundreds of
thousands of compounds to the targets of known structure. It is usually based on molecular docking. In molecular docking, a small
molecule is fitted into the active site of protein model and here, comparison of the 3D structure of small molecule with the putative
binding pocket is carried out. In the traditional HTS, the general ability of a ligand to bind, inhibit or allosterically alter the proteins
function is asserted experimentally, whereas in SB-vHTS selects the ligands that are predicted to bind to a specific binding site. To ensure
the feasibility of screening of large compound libraries within a finite time, limited conformational sampling of proteins and ligands is
used by SB-vHTS along with a simplified approximation of binding energy that can be computed rapidly (Becker et al., 2006).

Structure-based virtual screening

This is a computational approach for identifying potential drug candidates (hits) that are capable of binding to a drug target
(protein receptors, enzymes). This method involves quick searching of large libraries of chemical followed by docking of the hit
into a protein target and finally application of a scoring function for estimating the probability of binding affinity of drug
candidate with the protein target (Cheng et al., 2012). The most important advantage of this screening is that it enhances the hit
rate by considerably decreasing the number of compounds that are estimated experimentally for their activity and hence improves

Fig. 6 An overview of CADD drug designing/design pipeline. Adapted and modified from Guido, R.V., Glaucius Oliva, G., Andricopulo, A.D., 2008.
Virtual screening and its integration with modern drug design technologies. Current Medicinal Chemistry 15, 37–46.
 Small Molecule Drug Design 13

the success rate of the in vitro experiments. This method has been applied extensively in pharmaceutical companies and academic
groups for early-stage drug discovery.

Fragment-based lead discovery

This approach is based upon structure-activity relationships (SAR), obtained from NMR for identifying and optimizing the lead
(Bienstock, 2011). High purity, weak potency but effective binding, good aqueous solubility, (molecular weighto300, ClogPo3, number
of rotatable bonds, number of hydrogen bond donors and acceptors each should be o3) are the criteria for selecting the chemical
fragments (Congreve et al., 2003). Later, these fragments are either expanded or combined for producing a lead with a higher affinity.

In silico structure-based lead optimization

After the desired hits are identified through virtual screening, this method speeds up the search for optimized lead by delineating
the prediction about its pharmacological properties, thereby reducing the in vitro and in vivo experimental time.

ADMET modelling
This method, a common name for which is physiologically-based pharmacokinetic modelling is used in drug design and
development, and in assessing of toxicity threat evaluation and specifically predicts absorption, distribution, metabolism,
excretion and toxicology (ADMET) of drugs/compounds in humans. The ADMET parameters are based on the kinetics of the drug
exposure to tissues and how the body will react to them, influencing the performance and pharmacological activity of the
compound. Therefore, this method provides a key insight into the behaviour of a pharmaceutical compound within an organism.
This approach aids in the selection of compounds during the very early phases of drug thereby playing a crucial role in drug
discovery and development. This technique is cost- and time effective owing to a reduction in attrition of drugs during the pre-
clinical / clinical phase trials at a later stage.

Ligand-based drug designing

The existing knowledge of active compounds against the target is used to predict new chemical entities that present similar behaviour in
Ligand-based methods (Martin et al., 2002). Given a single known active molecule, a pharmacophore model can be derived from a library
of molecules to define the minimum necessary structural characteristics a molecule must possess in order to bind to the target of interest. A
fingerprint-based similarity search is usually used to compare the active molecule to the library as here, the molecules are represented as bit
strings which represent the presence or absence of predefined structural descriptors (Mishra and Siva-Prasad, 2011). In comparison,
targeting structural information to determine whether a new compound is likely to bind and interact with a receptor is the method that
structure-based methods rely on. No prior knowledge of active ligands is required in this method, which is a significant advantage (Kolb
et al., 2009). It is possible to design new ligands that can elicit a therapeutic effect from 3D structures. Therefore, the development of new
drugs through the discovery and optimization of the initial lead compound are greatly impacted by structure-based approaches.

Ligand-based virtual screening (LBVS)

Ligand-based virtual screening is based on the “similarity principle” according to which similar molecules tend to exhibit similar
biological properties. Scaffold hopping i.e., identification of iso-functional molecular structures with significantly different
molecular backbones is the usual objective when using LBVS. “Scaffold hopping” is also known as “leapfrogging”, “scaffold
searching” and “leap hopping” (Kalliokoki, 2010). These methods are usually helpful in drug repurposing, wherein new targets
and diseases are pursued for existing drug molecules.

Molecular descriptors
This is one of the simplest approaches in which the reference molecule/set of molecules are compared with a large library of
compounds at a very low cost on the basis of physicochemical properties descriptors, such as molecular weight, volume, geometry,
surface areas, atom types, dipole moment, polarizability, molar refractivity, octanol-water partition coefficient (log P), planar
structures, electronegativity, or solvation properties that are obtained from experimental measurements or theoretical models.
Molecules are represented by symbols for effective execution of the task (Prada-Graciaa et al., 2016).

Quantitative structure-activity relationship models (QSAR)

The mathematical relation between structural attributes and target response for a set of chemicals are explained by Quantitative
Structure-Activity Relationship models. Structural and/or property descriptors of compounds can also be correlated with their
biological activities using QSAR (Bernard et al., 2005). Through QSAR models we can correlate various features like rate constants,
binding sites affinities of ligands, inhibition constants and other biological activities, either with certain structural features (Free
Wilson analysis) or with atomic, molecular or group properties, such as lipophilicity, electronic, steric and polarizability, among
congeneric series of compounds. (Kubinyi, 1995). Hence, the success of QSAR is dependent on the choice of descriptors and the
ability to generate the appropriate mathematical relationship besides the quality of initial set of active/inactive compounds.

Pharmacophore modelling
More significant information can be drawn by employing various conformations of a range of ligands than just a single ligand
structure. A pharmacophore model of the receptor site can be generated with a sufficiently broad range of ligands. Pharmacophore
14 Small Molecule Drug Design

Fig. 7 Structures of Gefitinib and Tarceva and their analogs.

Fig. 8 (A) The sequence alignment of the target protein (FLT3) with the template structure in the DFG loop region. (B) Comparison of the DFG-in
and DFG-out FLT3 structures. Adapted and modified from Ke, Y.Y., Singh, V.K., Coumar, M.S., et al., 2015. Homology modeling of DFG-in FMS-
like tyrosine kinase 3 (FLT3) and structure-based virtual screening for inhibitor identification. Scientific Reports 5, 11702. Doi: 10.1038/srep11702.
 Small Molecule Drug Design 15

modelling of smaller, non-peptide molecules that might have improved stability and bioavailability over their peptide counter-
parts has resulted in successful outcomes so far (Nielsen et al., 1999).

Case Studies
Computer aided drug designing (CADD) for EGFR protein controlling lung cancer
Epidermal growth factor receptor (EGFR), a receptor tyrosine kinase plays an important role in tumour cell survival. Phos-
phorylated EGFR upon activation causes phosphorylation of downstream proteins that lead to changes in cell proliferation,
invasion, metastasis, and inhibition of apoptosis. Human EGFR was taken as a protein by Baskaran et al. (2012), and commer-
cially available drugs (such as Gemzar, Gefitinib, Tarceva) were taken as ligands. The drugs were docked to this receptor and
Gefitinib and Tarceva were chosen based on the energy values. To improve the binding efficiency and steric compatibility of these
two drugs, modifications were made to the probable functional groups which were interacting with the receptor molecule. Analogs
were prepared using ACD Chem Sketch in MOL format which were then converted to 3D structure using Weblab Viewer Lite, a 3D
modelling program. It was then docked using Vega ZZ docking software. Gefitinib Analog 2(281) and Tarceva Analog 7 had
significantly lower energy values and were found to be better than the conventional drugs available (Fig. 7).

Structure-based virtual screening (SBVS), for the identification of new chemotypes for FLT3 inhibition
FMS-like tyrosine kinase 3 (FLT3) is a type III tyrosine kinase receptor and is highly expressed in hematopoietic stem and
progenitor cells. FLT3 binds to extracellular domain which activates cytoplasmic tyrosine kinase activity leading to activation of

Fig. 9 Computer-aided drug design (CADD) strategy for FLT3 inhibitor identification. Adapted and modified from Ke, Y.Y., Singh, V.K., Coumar,
M.S., et al., 2015. Homology modeling of DFG-in FMS-like tyrosine kinase 3 (FLT3) and structure-based virtual screening for inhibitor
identification. Scientific Reports 5, 11702. Doi: 10.1038/srep11702.
16 Small Molecule Drug Design

downstream cellular signalling which is essential for proliferation. Based on the position of the Phe residue of the DFG motif, the
inhibitors are named Type I or Type II inhibitors. In a DFG-in conformation, the Phe residue is oriented outside the ATP binding
site while in a DFG-out conformation; the Phe residue is oriented inside the ATP binding site. Inhibitors that bind to the DFG-in
conformation are termed type-I inhibitors, and those that bind to the DFG-out conformation are referred to as type–II inhibitors.
In Fig. 8, Ke et al. (2015), superimposed the FLT3-modeled (DFG-in) structure on the 1RJB (DFG-out) template structure to depict
the structural differences between the DFG-in and DFG-out conformations. Since type-I inhibitors bind strongly to FLT3-ITD-
mutated kinase, they are proven to be more effective against treatment of acute myeloid leukemia.
Homology modelling (HM) of the DFG-in FLT3 structure identifies known DFG-in (SU11248, CEP-701, and PKC-412) and
DFG-out (Sorafenib, ABT-869 and AC220) FLT3 inhibitors using docking studies (citation of the study). SBVS of an HTS library of
125,000 compounds was done with the modeled structure. Out of the 97 top scoring compounds, two hits (BPR056, IC50 ¼ 2.3
and BPR080, IC50¼ 10.7 mM) were identified. Based on molecular dynamics simulation and density functional theory calculations
BPR056 (MW: 325.32; cLogP: 2.48) was identified to interact with FLT3 in a stable manner and was studied to be chemically
optimizable to realize a drug-like lead in the future. The overall CADD strategy used in this study is shown in Fig. 9.
This superimposition reveals that the activation loop in the DFG-in structure flipped away from the active site which leads to the
placement of the Phe830 group away from this site. In the DFG-out structure, the activation loop overturned towards the active site,
resulting in the placement of Phe830 in the active site. The different arrangements of the activation loop, particularly in regards to the
placement of the Phe830 group, have profound implications for the binding of different types of inhibitors to FLT3 kinase.
More than 40% inhibition at a 10 mM concentration was observed for BPR056 and BPR080, out of 97 screened compounds (Table 2).
In order to design better FLT3 inhibitors it is important to understand the binding modes of these two high scoring poses in the
modeled structure. Both the molecules were observed to bind at the ATP binding site by forming Hydrogen bonds with the

Table 2 FLT3 kinase inhibition profiles, docking score and binding energies of the hits identified from the in-house HTS database. PKC412 and
sorafenib are used as reference compounds for the comparison

Compound FLT3 IC50 (nM) Docking score Predicted binding energy (kcal/mol)

DFG-inDFG-outDFG-in DFG-out

2300
355
276
23.04 16.41

10,700
350
314
22.92 19.82

Midostaurin (PKC412) 37
269
165
34.21 19.13

Sorafenib (BAY43-9006) 102
183
300
7.06
12.32

Adapted and modified from Ke, Y.Y., Singh, V.K., Coumar, M.S., et al., 2015. Homology modeling of DFG-in FMS-like tyrosine kinase 3 (FLT3) and structure-based virtual screening
for inhibitor identification. Scientific Reports 5, 11702. Doi: 10.1038/srep11702.
 Small Molecule Drug Design 17

hinge-region Cys694 residue. A hinge-region H-bond interaction between an inhibitor and kinase is an essential component in
inhibitory activity and are common in several inhibitor–kinase complex structures.
DFG and FLT3 inhibitor complex was subjected to 20 ns using the Gromacs MD suite. The root mean square deviation showed
that the both cases found to be at equilibrium from initial position and remained stable after 10 ns. In addition RMSF values of
complex were also determined and it showed that the major differences occurs between the both complexes occurred in the loop
region and more fluctuations in the FLT3-BPR080 interacting complex (Ke et al., 2015). Detailed docking, DFT calculations
and 20-ns MD simulations of the new hits in the DFG-in FLT3-modeled structure suggest that the interaction between BPR056
(MW: 325.32; cLogP: 2.48) and FLT3 is stable and could, in the future, be chemically optimized to realize a drug-like lead.

Fragment-based drug design to target the EphA4 kinase domain

The first approved drug that was developed via fragment-based approach is Vemurafenib (4, Fig. 10), a kinase inhibitor of the B-
Raf V600E oncogenic mutant. High content screening (HCS), a type of phenotypical screening, followed by X-ray crystallography
yielded an unselective 7-Azaindole fragment (Figure 12) as hit fragment, which was optimized via 2 into the selective B-Raf
inhibitor PLX472077 (3) to end up with Vemurafenib (PLX4032, 4) (Bollag et al., 2010).

Limitations of CADD

Despite the emergence of the numerous above-mentioned pioneering approaches, drug design and development is still an inherently
risky business where the input costs are high and the success rate is low. Generally, only one in 1000 lead compounds reaches phase 1
clinical trials and only one in five drugs make it from phase 1 trials into the marketplace (Wishart, 2006). Some procedures concerning
Computer Aided Drug Designing are time consuming, especially while looking for a proper lead component (Jorgensen, 2004). The
current molecular docking algorithms do not estimate the absolute energy associated with the intermolecular interaction with
satisfactory accuracy as most of the available scoring functions are classical approximations of events ruled by quantum mechanics.
Further, there is lack of accurate experimental data that restricts further advancement of CADD (Bharath et al., 2011). Hence, new
experimental or computational tools and scientific approaches to identify correlations between the nature and structure-based
properties of the drug and of its safety and efficacy in the human body (pharmacovigilance-based) are in vital need of improvement so

Fig. 10 Optimization of the non-selective 7-Azaindole fragment (24) into Vemurafenib (PLX4032, 27).
18 Small Molecule Drug Design

potentially problematic drug leads could be identified at the early stages in their development. This will improve the public health and
provide a safe, effective and rational use of medicines and their development.

Conclusion & Future Perspectives

CADD has now become an indispensable tool in the long process of drug discovery and development. It also provides options for
understanding chemical systems in different ways, yielding information that is not easy to obtain in laboratory analysis, with
considerably less cost and effort than experiments. Initially, CADD had a rocky perception in the field of drug development, and
perhaps some over-hyping of its promises, as is present in the initial stages of almost any new technology or development. Today,
one can say that the discipline of computational medicinal chemistry has begun to mature and is a routinely used component of
modern drug discovery process. Mastering different kinds of CADD approaches and their software and utilizing all computational
resources that are valuable for drug design are certainly essential for becoming a successful computational medicinal chemist in
today’s world. In addition, having skills in one or more programming languages, such as Python or JAVA will help smooth routine
drug-design work. SBVs and LBVs are also very likely to become routine in drug-discovery projects if they are not considered to
have already done so. The use of more accurate methods like MD and QM, continue to grow. In conclusion, CADD is beneficial for
pharmaceutical development in the areas of prediction of 3D structures, design of compounds, prediction of druggability, in silico
ADMET prediction however, it must be realised that computational predictions need to be integrated with experimental
approaches for successful drug discovery and development.

References

Abrams, C.F., Vanden-Eijnden, E., 2010. Large-scale conformational sampling of proteins using temperature-accelerated molecular dynamics. Proceedings of National Academy
of Sciences of the United States of America 107, 4961–4966.
Allen, M.P., Tildesley, D.J., 1989. Computer Simulation of Liquids. Oxford, UK: Oxford Science Publications, p. 385.
Baskaran, C., Bai, V.R., Kumaran, K., Kumar, K.M., 2012. Computer aided drug designing (CADD) for EGFR protein Controlling lung cancer. Annals of Biological Research 3
(4), 1815–1820.
Becker, O.M., Dhanoa, D.S., Marantz, Y., et al., 2006. An integrated in silico 3D model-driven discovery of a novel, potent, and selective amidosulfonamide 5-HT1A agonist
(PRX-00023) for the treatment of anxiety and depression. Journal of Medicinal Chemistry 49, 3116–3135.
Bernard, D., Coop, A., MacKerell Jr., A.D., 2005. Computer-aided drug design: Structure-activity relationships of delta opioid ligands. Drug Design Review 2, 277291.
Bharath, E.N., Manjula, S.N., Vijaychand, A., 2011. In silico drug design tool for overcoming the innovation deficit in the drug discovery process. International Journal of
Pharmacy and Pharmaceutical Sciences 3 (2), 1–5.
Bienstock, J.R., 2011. Overview: Fragment-based drug design. ACS Symposium Series 1076, 1–26.
Blaney, F., 1990. Molecular modelling in the pharmaceutical industry. Chemistry and Industry (London) 23, 791–794.
Bohm, H.J., Stahl, M., 2000. Structure-based library design: Molecular modeling merges with combinatorial chemistry. Current Opinion in Chemical Biology 4, 283–286.
Bollag, G., Hirth, P., Tsai, J., et al., 2010. Clinical efficacy of a RAF inhibitor needs broad target blockade in BRAF-mutant melanoma. Nature 467, 596–599.
Buchan, D.W., Ward, S.M., Lobley, A.E., et al., 2010. Protein annotation and modelling servers at University College London. Nucleic Acids Research 38, 563–568.
Bugg, C.E., Carson, W.M., Montgomery, J.A., 1993. Drugs by design. Scientific American Magazine. 60–66.
Cheng, T., Li, Q., Zhou, Z., Wang, Y., Bryant, S.H., 2012. Structure-based virtual screening for drug discovery: A problem-centric review. American Association of
Pharmaceutical Scientists (AAPS) Journal 14, 133–141.
Congreve, M., Carr, R., Murray, C., Jhoti, H., 2003. A ‘rule of three’ for fragment based lead discovery? Drug Discovery Today 8, 876–877.
Congreve, M., Murray, C.W., Blundell, T.L., 2005. Structural biology and drug discovery. Drug Discovery Today 10 (13), 895–907.
Dougueta, D., Thoreaua, E., Grassy, G., 2000. A genetic algorithm for the automated generation of small organic molecules: Drug design using an evolutionary algorithm.
Journal of Computer-Aided Molecular Design 14, 449–466.
Grubmüller, H., 1995. Predicting slow structural transitions in macromolecular systems: Conformational flooding. Physical Review E. Statistical, Physics, Plasmas, Fluids and
Related Interdisciplinary Topics 52, 2893–2906.
Guido, R.V., Glaucius Oliva, G., Andricopulo, A.D., 2008. Virtual screening and its integration with modern drug design technologies. Current Medicinal Chemistry 15, 37–46.
Huey, R., Morris, G.M., Olson, A.J., Goodsell, D.S., 2007. A semiempirical free energy force field with charge-based desolvation. Journal of Computational Chemistry 28,
1145–1152.
Jorgensen, W.L., 2010. Drug discovery: Pulled from a protein’s embrace. Nature 466, 42–43.
Jorgensen, W.L., 2004. The many roles of computation in drug discovery. Science 303, 1813–1817.
Kalliokoki, Tumo., 2010. Accelerating three dimensional virtual screening. Dissertations In Health Sciences.
Kalyaanamoorthy, S., Chen, Y.P., 2011. Structure-based drug design to augment hit discovery. Drug Discovery Today 16, 831–839.
Kapetanovic, I.M., 2008. Computer-aided drug discovery and development (CADDD) – In silico-chemical biological approach. Chemico-Biological Interactions 171, 165–176.
Ke, Y.Y., Singh, V.K., Coumar, M.S., et al., 2015. Homology modeling of DFG-in FMS-like tyrosine kinase 3 (FLT3) and structure-based virtual screening for inhibitor
identification. Scientific Reports 5, 11702. doi:10.1038/srep11702.
Kitchen, D.B., Decornez, H., Furr, J.R., Bajorath, J., 2004. Docking and scoring in virtual screening for drug discovery: Methods and applications. Nature Reviews Drug
Discovery 3, 935–949.
Klebe, G., 2000. Recent developments in structure-based drug design. Journal of Molecular Medicine 78, 269–281.
Kolb, P., Ferreira, R.S., Irwin, J.J., Shoichet, B.K., 2009. Docking and chemoinformatic screens for new ligands and targets. Current Opinion in Biotechnology 20, 429–436.
Kubinyi, H., 1995. in Burger’s Medicinal Chemistry. Wolff, M.E. (Ed.), vol. 1. John Wiley & Sons, pp. 497–571.
Kuhn, P., Wilson, K., Patch, M.G., Stevens, R.C., 2002. The genesis of high-throughput structure-based drug discovery using protein crystallography. Current Opinion in
Chemical Biology 6 (5), 704–710.
Lewis, R.A., 2005. A general method for exploiting QSAR models in lead optimization. Journal of Medicinal Chemistry 48 (5), 1638–1648.
Liu, M., Wang, S.M., 1999. MCDOCK: A Monte Carlo simulation approach to the molecular docking problem. Journal of Computer-Aided Molecular Design 13, 435–451.
Lombardino, J.G., Lowe, J.A., 2004. The role of the medicinal chemist in drug discovery — then and now. Nature Reviews Drug Discovery 3, 853–862.
Maithri, G., Manasa, B., Vani, S.S., Narendra, A., Harshita, T., 2016. Computational drug design and molecular dynamic studies – A review. International Journal of Biomedical
Data Mining. 1–7.
 Small Molecule Drug Design 19

Mallipeddi, P.L., Kumar, G., White, S.W., Webb, T.R., 2014. Recent advances in computer-aided drug design as applied to anti-influenza drug discovery. Current Topics in
Medicinal Chemistry 14, 1875–1889.
Marrone, T.J., Briggs, J.M., McCammon, J.A., 1997. Structure-based drug design: Computational advances. Annual Review of Pharmacology and Toxicology 37, 71–90.
Martin, Y.C., Kofron, J.L., Linda, M., Traphagen, L.M., 2002. Do Structurally Similar Molecules Have Similar Biological Activity? Journal of Medicinal Chemistry 45,
4350–4358.
Martí-Renom, M.A., Stuart, A.C., Fiser, A., et al., 2000. Comparative protein structure modeling of genes and genomes. Annual Review of Biophysics and Biomolecular
Structure 29, 291–325.
Martis, E.A., Radhakrishnan, R., Badve, R.R., 2011. High-throughput screening: The hits and leads of drug discovery – An overview. Journal of Applied Pharmaceutical Science
1 (1), 2–10.
Mishra, V., Siva-Prasad, CV., 2011. Ligand based virtual screening to find novel inhibitors against plant toxin Ricin by using the ZINC database. Bioinformation 7, 46–51.
Moore, D.J., West, A.B., Dawson, V.L., Dawson, T.M., 2005. Molecular pathophysiology of Parkinson's disease. Annual Review of Neuroscience 28, 57–87.
Morris, G.M., Goodsell, D.S., Halliday, R.S., et al., 1998. Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function. Journal of
Computational Chemistry 19, 1639–1662.
Nielsen, K.J., Adarns, D., Thomas, L., et al., 1999. Structure-activity relationships of omega-conotoxins MVIIA, MVIIC and 14 loop splice hybrids at N and P/Q-type calcium
channels. Journal of Molecular Biology 289 (5), 1405–1421.
Ou-Yang, S.S., Lu, J.Y., Kong, X.Q., et al., 2012. Computational drug discovery. Acta Pharmacologica Sinica 33, 1131–1140.
Prada-Graciaa, D., Huerta-Yépezb, S., Moreno-Vargasb, L.M., 2016. Application of computational methods for anticancer drug discovery, design, and optimization. Boletin
Medico Del Hospital Infantil De Mexico 73 (6), 411–423.
Schlitter, J., Engels, M., Krüger, P., 1994. Targeted molecular dynamics: A new approach for searching pathways of conformational transitions. Journal of Molecular Graphics
12, 84–89.
Schneider, G., Bohm, H.J., 2002. Virtual screening and fast automated docking methods. Drug Discovery Today 7, 64–70.
Shoichet, B.K., 2004. Virtual screening of chemical libraries. Nature 432, 862–865.
Sugita, Y., Okamoto, Y., 1999. Replica-exchange molecular dynamics method for protein folding. Chemistry Physics Letters 314, 141–151.
Veselovsky, A.V., Ivanov, A.S., 2003. Strategy of computer-aided drug design. Current Drug Targets-Infectious Disorders 3, 33–40.
Walters, W.P., Stahl, M.T., Murcko, M.A., 1998. Virtual screening – An overview. Drug Discovery Today 3, 160–178.
Wang, R.X., Lai, L.H., Wang, S.M., 2002. Further development and validation of empirical scoring functions for structure based binding affinity prediction. Journal of Computer-
Aided Molecular Design 16, 11–26.
Wishart, D.S., 2006. Metabolomics for drug discovery, development and monitoring. Business Briefing: Future Drug Discovery. 1–3.
Zhao, H., Huang, D., Caflisch, A., 2012. Discovery of tyrosine kinase inhibitors by docking into an inactive kinase conformation generated by molecular dynamics.
ChemMedChem 7 (11), 1983–1990.

Further Reading
Arumugasamy, K., Tripathi, S.K., Singh, P., Singh, S.K., 2016. Protein-protein interaction for the de novo design of cyclin-dependent kinase peptide inhibitors. In: Orzáez, M.,
Sancho Medina, M., Pérez-Payá, E. (Eds.), Cyclin-Dependent Kinase (CDK) Inhibitors. Methods in Molecular Biology, vol. 1336. New York, NY: Humana Press.(ISBN: 978-
1-4939-2926-9).
Baig, H.M., Ahmad, K., Roy, S., et al., 2016. Computer aided drug design: Success and limitations. Current Pharmaceutical Design 22 (5), 572–581.
Barril, X., 2017. Computer-aided drug design: Time to play with novel chemical matter. Expert Opinion on Drug Discovery 12 (10), 977–980.
Brooijmans, N., Kuntz, I.D., 2003. Molecular recognition and docking algorithms. Annual Review of Biophysics and Biomolecular Structure 32, 335–373.
Bursulaya, B.D., Totrov, M., Abagyan, R., Brooks 3rd, C.L., 2003. Comparative study of several algorithms for flexible ligand docking. Journal of Computer-Aided Molecular
Design 17 (11), 755–763.
Friesner, R.A., Banks, J.L., Murphy, R.B., et al., 2004. Glide: A new approach for rapid, accurate docking and scoring. 1 Method and assessment of docking accuracy. Journal
of Medicinal Chemistry 47 (7), 1739–1749.
Godwin, R.C., Melvin, R., Salsbury, F.R., 2015. Molecular dynamics simulations and computer-aided drug discovery. In: Zhang, W. (Ed.), Computer-Aided Drug Discovery.
Methods in Pharmacology and Toxicology. New York, NY: Humana Press. ISBN: 978-1-4939-3521-5.
Liao, C., Sitzmann, M., Pugliese, A., Nicklaus, M.C., 2011. Software and resources for computational medicinal chemistry. Future Medicinal Chemistry 3 (8), 1057–1085.
Muegge, I., Bergner, A., Kriegl, J.M., 2017. Computer-aided drug design at Boehringer Ingelheim. Journal of Computer-Aided Molecular Design 31 (3), 275–285.
Talele, T.T., Khedkar, S.A., Rigby, A.C., 2010. Successful applications of computer aided drug discovery: Moving drugs from concept to the clinic. Current Topics in Medicinal
Chemistry 10 (1), 127–141.
Wathieu, H., Issa, N.T., Stephen, W.B., Dakshanamurthy, S., 2016. Harnessing polypharmacology with computer-aided drug design and systems biology. Current
Pharmaceutical Design 22 (21), 3097–3108.
Wong, Y.H., Chiu, C.C., Lin, C.L., et al., 2016. A new era for cancer target therapies: Applying systems biology and computer-aided drug design to cancer therapies. Current
Pharmaceutical Biotechnology 17 (14), 1246–1267.
Yoshifumi, F., Tadaaki, M., Kiyotaka, M., et al., 2016. Miscellaneous topics in computer-aided drug design: Synthetic accessibility and GPU computing, and other topics.
Current Pharmaceutical Design 22 (23), 3555–3568.

Relevant Websites

http://chem.sis.nlm.nih.gov/chemidplus
ChemIDplus.
www.ebi.ac.uk/Tools/sss/psiblast
European Bioinformatics Institute. PSI-BLAST.
www.moldiscovery.com
Molecular Discovery.
www.ncbi.nlm.nih.gov
National Center for Biotechnology Information.
http://dtp.nci.nih.gov/webdata.html
NCI discovery services.
20 Small Molecule Drug Design

http://cactus.nci.nih.gov/chemical/structure
NCI/CADD Chemical Identifier Resolver.
http://195.178.207.233/PASS2008/en/index.html
Prediction of activity spectra for substances.
www.pdb.org
Protein Data Bank.
http://predictioncenter.org
Protein Structure Prediction Center.
www.chemspider.com
Royal Society of Chemistry. ChemSpider.
www.schrodinger.com
Schrödinger Inc.
www.simulations-plus.com
Simulations Plus, Inc.
http://thomsonreuters.com/products_services/science/science_products/a-z/world_drug_index/
Thomson Reuters World Drug Index.
http://en.wikipedia.org/wiki/Molecular_dynamics#Major_software_for_MD_simulations
Wikipedia. MD simulation program list.