‘Metabolite-likeness’ for design and selection of Pharmaceutical

drug libraries

Ankit,

M.S.(Pharm.), 3rd Semester,

Dept. of Pharmacoinformatics,

NIPER, S.A.S. Nagar, Punjab.1

Outline

• Introduction

• Comparing drugs and library compounds to metabolites

• Metabolite-likeness curves

Compound set sources

Overview of protocol

Processing structures

Molecular Descriptors

Clustering and representative sets

• Dissimilar drugs

• Conclusion

• References2

Introduction

• Drug : A drug is, in the broadest of terms,

a chemical substance that has known biological effects

on humans or other animals.

• Hit molecule : A hit is usually defined as a molecule which

binds to the target, which has been identified to be important

in the disease of interest.

• Lead molecule : A molecule showing expected efficacy in

relevant in vitro pharmacological models, often also passed

the preliminary in vitro toxicological and ADME screening3

4

• The search for pharmaceutically active drugs with desirable

properties and negligible side effects can be considered as a

multi objective optimisation problem over an enormous search

space of ‘possible’ drugs.

• According to Opera et al., ‘lead structures exhibit, on the

average, less molecular complexity (less MW, less number of

rings and rotatable bonds), are less hydrophobic (lower cLogP

and LogD), and less druglike’ than actual drugs.

5

• To narrow the search space of chemically diverse candidate

compounds, cheminformatic methods are used to constrain

the compounds screened such that they tend to display ‘lead-

likeness’ or ‘drug-likeness’ (and even ‘CNS likeness’)

• Same concept used for promiscous drugs having multiple

intended targets or poly-pharmacology

6

• Various filters used for screening purposes :

1. Lipinki’s Rule of Five (Ro5)

2. Rule of Three (Ro3)

3. Veber rule

4. MDDR like rules

5. Ghose filter

6. BBB filter

7. QED (Quantitative estimation of drug likeness)

8. CMC-50 like rules

7

• Lipinski rules do not normally cover drugs derived from

natural products.

• These rules are biophysical rather than structural in nature.

• General descriptors are not entirely effective when drugs are

mainly transported by carriers.

8

• Metabolite or Endogenite : Metabolites are small

molecule components of primary metabolism and not the

products of the reaction of drugs with drug metabolising

enzymes

• Present drugs are more metabolite-like than the typical

contents of pharmaceutical screening libraries ???

• Endogenite-likeness might be useful filter in design and

analysis of pharmaceutical libraries for drug discovery

9

Comparing drugs and library

compounds to metabolites

• To assess the relationship between drugs and metabolites,

compared against a background of compounds of the kinds

that typically make up screening collections for hit discovery,

which we refer to as library compounds.

• Represent pre-drugs and can be considered as starting points

for drug discovery and development.

10

Class Source Compounds Total

Metabolite HMDB 2835 5333

Metabolite Palsson 806

Metabolite BioCyc 772

Metabolite BIGG 698

Metabolite Edinburgh 2048

Drug DrugBank 4152 7330

Drug KEGG Drug 4435

Pre-drug ZINC 62,390

Table 1 : Sources of drug, metabolite and pre-drug structures

11

Fig.1. Histograms for different simple molecular properties

12

Metabolite-likeness curves

• The similarity of drug and library compounds to metabolites

is assessed by calculating the Tanimoto distance to the closest

metabolite

• A variety of molecular descriptors were computed, and

similarities calculated using the Tanimoto coefficient

13

Molecular descriptors

• Extended connectivity fingerprints : operate by

identifying the substructural environment of each atom upto a

diameter of 4. Thus descriptor is set of substructures in a

molecule.

Similar molecules share more substructures than dissimilar

molecules.

• Path fingerprints : describe a compound by all paths

through the molecular graph upto length of 4.14

Using Pipeline Pilot implementation of a Daylight-like path

fingerprint.

Paths are not branched therefore represent the molecule

differently.

• E-state indices : captures the electronic and topological

prpperties of atoms via electronic interactions with

neighbouring atoms and by distance on the molecular

graph.

15

• MDL keys : are substructural features observed to be of

utility in retrieval tasks like database searching.

Of the full set of 960 useful substructures , the definitions of

166 were released as the MDL Public keys.

16

Clustering and representative sets

Descriptor space Drug Library

Connectivity fingerprint 5723 44,275

Path fingerprint 5835 44,318

E-state 6028 44,419

MDL Public Keys 5813 44,325

17

Table 2 : Sizes of representative sets in each of the descriptor spaces

Overview of protocol

18

Processing structures

• All compounds were ‘washed’ in Pipeline Pilot.

• Wahing involved :

isolation of the largest fragment in the structure

removal of salts

removal of hydrogens

Standardisation of stereochemical and charge information

Only compound with more than 3 atoms were considered

19

Fig.2. A comparison of drug and library distances to closest metabolites in

various molecular descriptor spaces

20

Dissimilar drugs

21

Conclusion

• It is desirable to develop filters that include transporters

effect also.

• With advanced system biology, there is need to move

towards more mechanistic approaches.

22

References

1. ‘Metabolite-likeness’ as a criterion in the design and selection of

pharmaceutical drug libraries. Paul D. Dobson, Yogendra Patel and

Douglas B. Kell. Drug Discovery Today Volume 14, Numbers 1/2

January 2009.

2. Navigating chemical space for biology and medicine. Lipinski, C.

and Hopkins, A .Nature 432, 855–861 2004.

3. Global mapping of pharmacological space. Paolini, G.V. et al. Nat. Biotechnol. 24, 805–815 2006.

23

24