International Journal of Pharmacy Research and Technology

2012, Volume 2, Issue 3, 01-05.

ISSN 2250 – 0944 (Online)
ISSN 2250 – 1150 (Print)
Review Article

Druggability of Genome

H. B. Patel*, S. K. Mody, A. B. Chukewar, C. M. Modi, G. B. Dudhatra, Avinash Kumar and Ratn DeepSingh
Department of Pharmacology & Toxicology, College of Veterinary Science & Animal Husbandry,
Sardarkrushinagar Dantiwada Agricultural University, Sardarkrushinagar-385506, Gujarat, India
*
Corresponding author: E-mail: drhitesh2002@rediffmail.com
Received: 21/2/2012, Revised: 19/05/2012 Accepted: 25/05/2012
ABSTRACT
The process of drug discovery in pre-genomic era was considered as very complex process requiring great deal of time and
money with high degree of unpredictability. The genomic research revolution has brought strategic shift in the concept of
drug discovery and development with availability of tool for early identification of molecular targets based on human
genomics. This approach has led to reduced rate of compound attrition and rejection, making process of drug discovery more
viable economically and to some extent predictable with respect to time frame. The criteria of compound which make the
compound fit for passing the process of drug discovery are studied and gene based targets are indentified, validated and
interpreted. This approach no doubt is at its infantile stage but has tremendous future potential to play key role in drug
discovery process. The present review describes the interrelationship between the druggability of genome and its impact on
drug discovery process.

Keywords: Druggability, Drug Discovery, Drug Targets, Genome

INTRODUCTION statistics of molecular drug targets and percentage of

In the beginning of 20th century, the pharmaceutical different molecular targets of current drugs are shown in
industry witnessed that too many compounds were Table 1 and Figure 1 respectively.
terminated in clinical development because of
unsatisfactory pharmacokinetic (PK) profile. The scenario What is druggability?
is somewhat still continues causing economical loss due to The druggability is defined as a presence of protein
unreasonable rate of drug failure. Medicinal chemists need folds that favour interactions with drug-like chemical
to address these problems during lead optimization and compounds [4]. An extension of Lipinski’s ‘Rule of Five’
therefore novel tools are in urgent need to assess the to protein targets that can bind, such ‘drug-like’ compounds
relationship between structure of compound and and therefore are thought to be amenable to modulation by
pharmacokinetic prospect. Search for an understanding of compounds with oral bioavailability has led to the terms
what is responsible for compound attrition has led to the like ‘druggable protein’ [2] and ‘druggable genome’ [4].
development of criteria which are characteristic for Proteins lacking these structural features might have
compounds that successfully pass through the development interesting biological properties, but are unlikely to be
process [1]. Such compounds have been called ‘druggable’ readily amenable to pharmaceutical modulation. The
or ‘drug-like’ [2, 3]. problem can be visualised as doors and keyholes. A door
The introduction of Lipinski’s ‘Rule of Five’ (RO5) (protein) might control access to an interesting pathway,
has initiated a profound shift in the thinking paradigm of but if it does not have the appropriate keyhole (druggable
medicinal chemists. Understanding the difference between domain) it cannot be opened. Analysis of druggability, so to
biologically active small molecules and drugs became a speak, is the analysis of keyholes [5].
priority in the drug discovery process, and the importance
of addressing pharmacokinetic properties early during lead Table: 1. Molecular drug targets [26, 27]
optimization is a clear result. These concepts of ‘drug- No. of molecular
Class of drug target
likeness’ and ‘druggability’ have been extended to proteins targets
and genes for target identification and selection [24, 25]. Targets of approved NMEs (Human
How should these concepts be integrated practically into 301
and anti-infectives)
the drug discovery process? This review summarizes the Human Targets of approved NMEs 238
recent advancement in the field and examines the Human Targets of approved NCEs 170
usefulness of druggability of compound and genome and its Targets of approved biological 59
prospects in drug discovery and development process. Human Targets of approved antibodies 15
Whole sequence of the human genome with high-
quality annotation is available; it is a good time to have a HISTORY
fresh look at the druggable genome. Analysis suggests a In 2002, Hopkins and Groom introduced the concept
druggable gene count can be between 2000 and 3000, in of the ‘druggable genome’ [4]. Their purpose was to
general agreement with previous estimates (3000). identify the limited set of molecular targets for which
However, there is evidence for a significant shift in the commercially viable, oral compounds can be developed.
contribution of the major target families, with fewer Because such targets are expected to bind RO5-compliant
rhodopsin-like GPCR (G-protein Coupled Receptor) and compounds, they analyzed databases and used
protein kinases, and more proteases than expected. The

IJPRT | July – September | 1

 
 Patel et al / International Journal of Pharmacy Research & Technology 2012 2(3) 01-05 

computational methods to identify all proteins belonging to families and taking into account that, by necessity, a drug
families which have at least one member that has target needs to have the potential to be disease-modifying,
successfully been targeted by drug-like molecules. they estimated that the human genome encodes 600–1500
Assuming that druggability is shared among protein targets for small-molecule drugs [5].

Figure: 1. Molecular target of current drugs [26, 27]

Features of druggability drug discovery process and it produces a large number of

(1) It is also important to keep in mind the distinction false positives and promiscuous hits which complicate the
between druggability of a protein, and its actual analysis.
qualities as a drug target [4]. Another method has recently been developed by
(2) Many proteins are druggable according to their researchers from Abbott. Using 2-D heteronuclear-NMR
structure, but modulating their biological function may they studied the interactions of 10000 lead-like or
not provide any therapeutic benefit; not every door fragment- like compounds with protein surfaces. This
with a keyhole leads to a desirable place. approach has the advantage that it samples a large fraction
(3) The actual drug targets are the subset of druggable of chemical space (even though the size of the library is
proteins which possess both structural and functional small) and yields more reliable data than conventional high-
features of druggability, and their only real validation throughput screening. Furthermore, such an NMR based
comes with successful clinical use. analysis of druggability could be performed with limited
(4) The numbers of protein target for commercially viable resources and relatively early in the drug discovery process.
compounds are considerably smaller than originally Most importantly, an analysis of the NMR data has led to
thought. the development of ‘druggability-indices’ that can be used
(5) The size of the Pharmaceutically Tractable Genome, a for the computational assessment of proteins with known
term that was suggested for genes that can be targeted structure [7].
by small molecular weight compounds, antibodies and
recombinant therapeutic proteins [6] is considerably Table: 2. Druggable genome predictions [26, 27]
larger. No. of molecular
(6) Multimeric protein complexes and successful Druggability prediction method
targets
promiscuous drugs acting on more than one target do Targets of approved NCEs 170
not allow to immediately generalizing from gene to Sequence homology to NCE drug
drug target. When discussing the merits of a druggable 945
targets
genome annotation, it is very important to keep all Targets of chemical leads with
these limitations in mind. activities (binding affinities) below 707
10 µM
Steps in druggability Targets of Ro5 chemical leads with
A first step towards a more reliable way to assess the activities (binding affinities <= 10 587
druggability of individual proteins is the identification of µM)
binding sites for drug-like molecules [7]. Kellenberger and Sequence homology to targets with
co-workers have created an annotated database of ligand 2921
chemical leads
binding sites extracted from several experimental structure Feature-based druggability sequence
databases [8]. Such data can be used to derive rules or 2325
probability prediction
training sets for the computational identification of binding Structured-based prediction 427
pockets. Many algorithms [9] are available for this purpose Sequence homology to proteins
and in general they have been successfully applied in predicted druggable by structure- 3541
identifying true ligand binding pockets on the surface of based method
proteins [8]. Predicted Druggbable Genome (high
The second step of quantitatively assessing the 3505
confidence)
druggability of the identified pockets is more challenging. Human Genome 23000
An obvious approach is to screen a large library of drug-
like compounds pharmaceutical compound collection or a How precisely can we define the druggable genome?
chemical genetics library [10] and assess the resulting hits. Over time, the estimated numbers of potential targets are
Unfortunately, this approach has three significant now converging at around 3000 druggable loci [4, 11, 12].
drawbacks: it is very expensive, applied rather late in the

2 | IJPRT | July – September

 Patel et al / International Journal of Pharmacy Research & Technology 2012 2(3) 01-05 

Main parameters influencing the count are the coverage of assembly covers 99% of the euchromatic genome in high
sequence databases, the tools used for sequence annotation, quality sequence, with only 341 gaps remaining. Thus,
structural information about tractable folds, bioinformatics current estimates should only miss 1% of potential targets
tools, and biological information about protein function. due to lack of database representation. The current status of
One key parameter, the genome sequence itself, has druggable genome prediction is depicted in Table 2 and
stabilised now. While the first working draft sequence Figure 2, 3.
covered only 90% of the genome, the preliminary final

Figure: 2. Gene family distributions of predicted druggable genome [27]

Figure: 3. Predicting protein druggability [7]

The results of druggability indices derived from 1096 Non-redundant human proteins, in which the percentage of
members of selected target classes that contain a druggable binding site (‘percent druggable’) is plotted. In this
preliminary analysis, no attempt was made to differentiate the known active or ligand-binding site, nor was
conformational flexibility taken into account. Overall, 35% of the targets in this set contained at least one druggable
binding site. Target classes depicted in green have a higher than average percentage of members with druggable
binding sites, whereas those depicted in red have a lower than average percentage. The horizontal gray bar
represents the average and standard deviation for the entire dataset

Is the power shifting between the ruling families? Obtaining an exact count of rhodopsin-like GPCR
The two largest families, protein kinases and drug targets is cumbersome, as hundreds of closely related
Rhodopsin-like GPCR, are still topping the league table, sensory receptors inflate the automated domain count, but
but are smaller than expected. IPRO and PFAM predictions are not likely to have any therapeutic potential. Near about
are very close to put the count for kinases at 480-500, down three hundred non-sensory members of the family have
more than 20% from earlier predictions and in close been identified, this is in agreement with the IUPHAR
agreement with the detailed annotation of the “kinome” receptor database [14] and other recent publications [15].
(518) [13]. Table 3 is showing comparison of the druggable Both predicting under 300 druggable members of this
genomes of selected eukaryotes. family.

IJPRT | July – September | 3

 Patel et al / International Journal of Pharmacy Research & Technology 2012 2(3) 01-05 

Table: 3. Comparison of the druggable genomes of selected eukaryotes [4]

Homo Drosophila Caenorhabditis Saccharomyces
sapiens melanogaster elegans cerevisiae
Total number of predicted genes ~30,000 13,601 18,424 6,241
Number of proteins in proteome 21,688 13,849 17,946 6,127
Number of estimated druggable targets 3,051 1,714 2,267 508
Percentage that are predicted druggable
~10–14% 12% 12% 8%
targets

These receptors, very many of them linked to disease- CONCLUSIONS

relevant physiology, are still most likely to be the richest What do these numbers tell us, and how stable can we
opportunity for drug discovery in the short- to mid-term, expect them to be? As discussed above, this approach to
but the lower number suggests that the area might soon be annotating the druggable genome estimates the potential
saturated by drug discovery efforts. The counts for non- maximum size of the playing field for current small
rhodopsin GPCR, ion channels, transporters, nuclear molecule drug design. It does not consider biological
hormone receptors and cytochromes are largely unchanged, (where the rules are harder to define), RNA (which is not
indicating that these smaller families are annotated in great based on protein structure), or future breakthroughs in
detail already. The most important area that consistently medicinal chemistry or biology (no crystal ball available).
appears larger than previously expected is the proteases. It is to be keep in mind that these simple sequence
Even the most stringent of counts (PFAM-CCDS), which is comparisons do not allow any immediate far-reaching
likely to be an underestimate, yields ~230 putative proteins. conclusions about the biological function of a protein.
Applying the IPRO models increases the count to ~380. However, sequence-based protein structure prediction
The higher numbers are consistent with recent curated certainly pinpoints areas of research that will be highly
counts [16]. Which predict 553 proteases and related enriched for “real” drug targets. Investigating the function
proteins (although they include additional domain of orphan receptors has been very successful in identifying
signatures that are not defined as druggable). It appears that unexpected novel players in important physiological
the druggable protease space might be at least as large as pathways [17, 18] and the druggable genome will further
the rhodopsin-like receptors. The heterogeneous group of guide experimental efforts to understand the biology of
druggable enzyme families seems to be of similar size as other potential drug targets.
previously predicted, but due to its complexity a detailed The RO5 and its extensions have been useful tools to
validation of this observation is very tedious. If it is generate awareness about the importance of PK parameters
assumed that the numbers for kinases, GPCR and smaller for development. In addition, this concept has led to the
target families will remain stable in the optimistic scenario, realization that there may be whole families of proteins for
the larger number of proteases compensates for the which it is extremely challenging to design compounds
reduction of other families, and arrives at just over 3000 with good oral bioavailability.
targets, the same total as Hopkins and Groom in 2002. The
conservative count uses the validated protease subfamilies REFERENCES
and high-stringency predictions for enzymes and other 1. C A Lipinski, Filtering in drug discovery, Ann. Rep.
target classes only, yielding a total of around 2200 Comp. Chem. 1, 155-168 (2005).
druggable genes. 2. A L Hopkins and C R Groom, Target analysis: a priori
The available evidence suggests that qualitative assessment of druggability, Ernst. Schering Res.
druggability arguments are useful strategic tools; however, Found Workshop. 42, 11-17 (2003).
more accurate, quantitative assessments are needed, 3. A P Orth, S Batalov, M Perrone, S K Chanda,  The
especially for proteins with borderline druggability. promise of genomics to identify novel therapeutic
Moreover, the breakthroughs in small molecule disruption targets, Expert Opin. Ther. Targets. 8 (6), 587-596
of protein–protein interactions (e.g. p53–MDM2 [19] and (2004).
Bcl-2–Bax [20]) might lead to further expansion of the 4. A L Hopkins and C R Groom, The druggable genome,
‘druggable’ genome. Since its discovery, nuclear magnetic Nat. Rev. Drug Discov. 1 (9), 727-730 (2002).
resonance (NMR) has become the single most powerful 5. G Muller, Medicinal chemistry of target family-
form of spectroscopy in both chemistry and structural directed master keys, Drug Discov. Today. 8 (15),
biology. The dramatic technical advances over the past 10– 681-691 (2003).
15 years, which continue apace, have markedly increased 6. R Morphy, C Kay and Z Rankovic, From magic
the range of applications for NMR in the study of protein– bullets to designed multiple ligands, Drug Discov.
ligand interactions. These form the basis for its most Today. 9 (15), 641-651 (2004).
exciting uses in the drug discovery process, which range 7. P J Hajduk, J R Huth, and C Tse, Predicting protein
from the simple identification of whether a compound (or a druggability, Drug Discov. Today. 10 (23-24) 1675-
component of a mixture) binds to a given protein, through 1682 (2005).
to the determination of the full three-dimensional structure 8. E Kellenberger, P Muller, C Schalon, G Bret, N Foata
of the complex, with all the information this yields for and D Rognan, Sc-PDB: an annotated database of
structure-based drug design. This is a very attractive druggable binding sites from the protein data bank, J.
approach that can be used in the early stages of drug Chem. Inf. Model. 46, 717-727 (2006).
discovery and provides a solid basis for computational 9. J An, M Totrov and R Abagyan, Comprehensive
druggability estimations [21, 22, 23]. identification of druggable protein ligand binding
sites, Genome Inform. Ser. Workshop Genome
Inform. 15, 31-41 (2004).

4 | IJPRT | July – September

 Patel et al / International Journal of Pharmacy Research & Technology 2012 2(3) 01-05 

10. F Darvas, G Dorman, L G Puskas, A Bucsai and L 19. L T Vassilev, et al., In vivo activation of the p53
Urge, Chemical genomics for fast and integrated target pathway by small-molecule antagonists of MDM2,
identification and lead optimization, Med. Chem. Res. Science. 303, 844-848 (2004).
13, 643-659 (2004). 20. T Oltersdorf, et al., An inhibitor of Bcl-2 family
11. J Drews and S Ryser, The role of innovation in drug proteins induces regression of solid tumours, Nature.
development, Nat. Biotechnol. 15 (13), 1318-1319 435, 677-681 (2005).
(1997). 21. P J Hajduk, J R Huth and S W Fesik, Druggability
12. J Drews, Drug discovery: a historical perspective, indices for protein targets derived from NMR-based
Science. 287 (5460), 1960-1964 (2000). screening data, J. Med. Chem. 48, 2518-2525 (2005).
13. G Manning, D B Whyte, R Martinez, T Hunter and S 22. G C K Roberts, Applications of NMR in drug
Sudarsanam, The protein kinase complement of the discovery, Drug Discovery Today. 5 (6), 230-240
human genome, Science. 298 (5600), 1912-1934 (2000).
(2002). 23. M Betz, K Saxena and H Schwalbe, Biomolecular
14. S M Foord, T I Bonner, R R Neubig, et al., NMR: a chaperone to drug discovery, Current Opinion
International Union of Pharmacology. XLVI. G in Chemical Biology. 10, 219-225 (2006).
Protein-Coupled Receptor List, Pharmacol. Rev. 57, 24. T H Keller, A Pichota and Z Yin, A practical view of
279-288 (2005). druggability, Current Opinion in Chemical Biology.
15. R Fredriksson, M C Lagerstorm, L G Lundin and H B 10, 357-361 (2006).
Schioth, The G-protein-coupled receptors in the 25. C A Lipinski, F Lombardo, B W Dominy and P J
human genome form five main families. Phylogenetic Feeney, Experimental and computational approaches
analysis, paralogon groups, and fingerprints, Mol. to estimate solubility and permeability in drug
Pharmacol. 63 (6), 1256-1272 (2003). discovery and development settings, Adv. Drug Deliv.
16. X S Puente, L M Sanchez, C M Overall, C Lopez- Rev. 46, 3-26 (2001).
Otin, Human and mouse proteases: a comparative 26. J P Overington, B Al-Lazikani and A L Hopkins, How
genomic approach, Nat. Rev. Genet. 4 (7), 544-558 many drug targets are there? Nature Reviews. 5, 993-
(2003). 996 (2006).
17. S B Seminara, S Massager, E E Chatzidaki, et al., The 27. A L Hopkins, The druggable genome: how do we
GPR54 gene as a regulator of puberty, N. Engl. J. deliver on the promise of new drug targets? An
Med. 349 (17), 1614-1627 (2003). International perspective on pharmacogenetics: the
18. S Messager, E E Chatzidaki, D Ma, et al., Kisspeptin intersection between innovation, regulation and health
directly stimulates gonadotropin-releasing hormone delivery, Rome, Italy (2005).
release via G protein-coupled receptor 54, Proc. Natl.
Acad. Sci. USA. 102 (5), 1761-1766 (2005).

IJPRT | July – September | 5