De novo Molecular Design

Gisbert Schneider

• Publisher: Wiley-VCH
• Year: 2013
• Print ISBN:
  9783527334612
• eBook ISBN:
  9783527677030
• Edition: 1
• Topic: Medicine
• Subtopic: Pharmacology

Systematically examining current methods and strategies, this ready reference covers a wide range of molecular structures, from organic-chemical drugs to peptides, Proteins and nucleic acids, in line with emerging new drug classes derived from biomacromolecules.

A leader in the field and one of the pioneers of this young discipline has assembled here the most prominent experts from across the world to provide first-hand knowledge. While most of their methods and examples come from the area of pharmaceutical discovery and development, the approaches are equally applicable for chemical probes and diagnostics, pesticides, and any other molecule designed to interact with a biological system. Numerous images and screenshots illustrate the many examples and method descriptions.

With its broad and balanced coverage, this will be the firststop resource not only for medicinal chemists, biochemists and biotechnologists, but equally for bioinformaticians and molecular designers for many years to come.


From the content:

* Reaction-driven de novo design
* Adaptive methods in molecular design
* Design of ligands against multitarget profiles
* Free energy methods in ligand design
* Fragment-based de novo design
* Automated design of focused and target family-oriented compound libraries
* Molecular de novo design by nature-inspired computing
* 3D QSAR approaches to de novo drug design
* Bioisosteres in de novo design
* De novo design of peptides, proteins and nucleic acid structures, including RNA aptamers

and many more.

Chapter 1

De Novo Design: From Models to Molecules

Gisbert Schneider and Karl-Heinz Baringhaus

Form ever follows function, and this is the law.

Where function does not change, form does not change.

Louis Sullivan, American architect (1896) [1]

Innovative bioactive agents fuel sustained drug discovery and the development of new medicines. Future success in chemical biology and pharmaceutical research alike will fundamentally rely on the combination of advanced synthetic and analytical technologies that are embedded in a theoretical framework that provides a rationale for the interplay between chemical structure and biological effect. A driving role in this setting falls on leading edge concepts in computer-assisted molecular design, by providing access to a virtually infinite source of novel druglike compounds and guiding experimental screening campaigns. In this chapter, we present concepts and ideas for the representation of molecular structure, suggest predictive models of structure–activity relationships, and discuss approaches that have proved their usefulness and will contribute to future drug discovery by generating innovative bioactive agents. We also highlight some of the current prohibitive aspects of fully automated de novo design that will require attention for future methodological breakthroughs. This chapter provides an introduction to important pillars of de novo drug design, whereas the subsequent contributions presented in this book offer in-depth treatments of current trends, methods, and approaches together with numerous practical examples. We are confident that the reading will inspire.

1.1 Molecular Representation

Ever since the first atomic models of molecules have been conceived, scientists have used such models, and their associated concepts and language, to come up with innovative chemical agents that possess sought properties [2]. So far, we tend to think of a molecule in terms of sticks and balls when it comes to visualize chemical structure. No doubt, simplistic representations have their justification for describing certain aspects of molecular constitution, configuration, and conformation and provide an intuitive access to “molecular architecture” (Figure 1.1). However, they fall far short of relating functional aspects to the objects we recognize as molecules. In the end, it is the desired function we wish to get from a molecular structure. “Form follows function” – this credo of modern architecture and industrial design is equally valid for molecular design, in particular in medicinal chemistry and chemical biology striving for new chemical entities (NCEs) as biologically active lead compounds and eventually future drugs.

Figure 1.1 Atomic models of molecular structure as depicted in John Dalton's seminal book entitled A New System of Chemical Philosophy (1808). Panel (a) presents the “arbitrary signs chosen to represent the several chemical elements or ultimate particles.” Panel (b) might be considered as an early molecular design study, as it depicts Dalton's view of various arrangements of water molecules. Note the similarity between these archaic philosophies and contemporary molecular models.

Ideally, one would like to obtain a compound with a desired function directly from a design hypothesis, for example, a mathematical model that serves as a blueprint, without the need for exhaustive screening and meticulous optimization. In fact, de novo design means generating new molecules with desired properties “from scratch.” The concept of using transition functions that assign new states to objects, thereby observing emergent system properties [3, 4], has been well researched in fields such as complexity analysis, dynamical system, game theory, and systems biology [5]. In molecular design, we use models of the molecular world and expect a trustworthy model to correctly reflect aspects of the real world, so it can be used for predicting new molecules that possess the target property reflected in the model (Figure 1.2a). De novo design theory is tightly related to solving the inverse quantitative structure–activity relationship (SAR) problem or – to paraphrase from a philosophical point of view – finding the “Urbild,”¹ that is, the structural archetype associated with a molecular representation. In terms of mathematics, one tries to find an element x that is related to the value ξ:

. In molecular design, x is a molecular structure from the set of all compounds (usually referred to as chemical space) and ξ is the representation (descriptor) of x computed by function f [8]. Typically, the representation of a compound is a real numbered value or set of values (vector representation), although other, for example, symbolic forms of representations have been suggested [9]. It is essential to realize that the representation of a chemical structure is always uniquely defined by the mapping function f, while there may exist – if defined – many possibly infinite numbers of molecules that have the exact same descriptor values (Figure 1.2b). As a basic illustration of this important point, consider the total charge descriptor of a molecule containing N atoms, which is computed as

, where

is the partial charge of atom i. Accordingly, it is easy to determine the total charge for a given molecular structure, but it there may be numerous chemically feasible compounds featuring the same total charge.

Figure 1.2 (a) Models of chemical space. (Adapted from Ref. [4].) Molecules in chemical space (real world) are lumped into an equivalence class (dotted circle) according to a structure–activity relationship model. In computer-based molecular design, appropriate algorithms act as transition functions so that changes of model states are faithfully reflected in the adaptation of molecular structure and function. (b) Molecular representation and design. A function

transforms a molecular structure x to its corresponding molecular descriptor ξ. One may call x the “Urbild” of ξ. In molecular design applications, molecules are often mapped to numerical descriptor values by surjective functions, meaning that multiple elements of X might be turned into the same element of Y by applying f. This property of many molecular descriptor sets is exploited by de novo design, which aims at finding new molecules in X that can be mapped to pharmacologically meaningful representations.

Generally speaking, molecular de novo design aims at generating new compounds that can be mapped to well-defined, preferred representations, that is, sets of descriptor values that characterize compounds with the desired biological or pharmacological activity. The challenge hereby is twofold, namely to

1. define a set of mathematical functions that characterize compounds with desired properties (i.e., they belong to the same equivalence class), and

2. for a given molecular representation, find corresponding Urbild compounds.

Consequently, as a prerequisite for successful design, we need an adequate representation of molecular structures and their physicochemical properties to allow the extraction of features that are responsible for a certain compound property or pharmacological activity (=function). Ideally, we need to understand the behavior of a molecule in different environments (e.g., in solution and in complex with a receptor) over time. Consequent physical treatment of molecular properties and dynamics can in principle be achieved based on solutions of the Schrödinger equation (Eq. (1.1)).

1.1

where

is the Hamilton operator defining the operations that need to be performed with the set of wave functions

(psi) of the particles of a molecular system and E is the system's potential energy. Of note, the square of the absolute value of the wave function,

, may be interpreted as a probability density, thereby providing a probabilistic access to the rigid, finite “balls and sticks” of classical molecular models. The Schrödinger equationprovides a rigorous theoretical foundation for ab initio quantum chemical (QC) and quantum mechanical (QM) calculations, which are grounded on a solid physical and mathematical framework without the necessity for empirical values or heuristics. Such calculations represent the formally most accurate way of calculating states and energies of molecular systems, allowing an assessment of conformational preferences, chemical reactivity, interaction potential, and so on. The problem is, however, that exact solutions of the Schrödinger equation cannot be obtained for molecules that are more complex than H₂⁺, which currently renders druglike compounds with an average molecular mass of 300–500 Da out of reach. For such molecules of interest, approximations and generalizations are required that prohibit exact solutions to be found. For example, the Born–Oppenheimer approximation treats atom nuclei as fixed, and only the movement of electrons is considered. A further approximation is the Hartree–Fock method that is grounded on solving the Schrödinger equation for each electron of the molecular system individually, thereby leading to single-electron wave functions (orbitals). Semiempirical approximations resulted in the Hückel theory of molecular orbitals(MOs),...

Table of contents

• Cover
• Related Titles
• Title Page
• Copyright
• List of Contributors
• Foreword
• Preface
• Chapter 1: De Novo Design: From Models to Molecules
• Chapter 2: Coping with Complexity in Molecular Design
• Chapter 3: The Human Pocketome
• Chapter 4: Structure-Based De Novo Drug Design
• Chapter 5: De Novo Design by Fragment Growing and Docking
• Chapter 6: Hit and Lead Identification from Fragments
• Chapter 7: Pharmacophore-Based De Novo Design
• Chapter 8: 3D-QSAR Approaches to De Novo Drug Design
• Chapter 9: Ligand-Based Molecular Design Using Pseudoreceptors
• Chapter 10: Reaction-Driven De Novo Design: a Keystone for Automated Design of Target Family-Oriented Libraries
• Chapter 11: Multiobjective De Novo Design of Synthetically Accessible Compounds
• Chapter 12: De Novo Design of Ligands against Multitarget Profiles
• Chapter 13: Construction of Drug-Like Compounds by Markov Chains
• Chapter 14: Coping with Combinatorial Space in Molecular Design
• Chapter 15: Fragment-Based Design of Focused Compound Libraries
• Chapter 16: Free Energy Methods in Ligand Design
• Chapter 17: Bioisosteres in De Novo Design
• Chapter 18: Peptide Design by Nature-Inspired Algorithms
• Chapter 19: De Novo Computational Protein Design
• Chapter 20: De Novo Design of Nucleic Acid Structures
• Chapter 21: RNA Aptamer Design
• Index