Christos A. Nicolaou

The Proximal Lilly Collection: Mapping, Exploring and Exploiting Feasible Chemical Space

Venturing into the immensity of the small molecule universe to identify novel chemical structure is a much discussed objective of many methods proposed by the chemoinformatics community. To this end, numerous approaches using techniques from the fields of computational de novo design, virtual screening and reaction informatics, among others, have been proposed. Although in principle this objective is commendable, in practice there are several obstacles to useful exploitation of the chemical space. Prime among them are the sheer number of theoretically feasible compounds and the practical concern regarding the synthesizability of the chemical structures conceived using in silico methods. We present the Proximal Lilly Collection initiative implemented at Eli Lilly and Co. with the aims to (i) define the chemical space of small, drug-like compounds that could be synthesized using in-house resources and (ii) facilitate access to compounds in this large space for the purposes of ongoing drug discovery efforts. The implementation of PLC relies on coupling access to available synthetic knowledge and resources with chemo/reaction informatics techniques and tools developed for this purpose. We describe in detail the computational framework supporting this initiative and elaborate on the characteristics of the PLC virtual collection of compounds. As an example of the opportunities provided to drug discovery researchers by easy access to a large, realistically feasible virtual collection such as the PLC, we describe a recent application of the technology that led to the discovery of selective kinase inhibitors.

Molecular library design using multi-objective optimization methods.

Advancements in combinatorial chemistry and high-throughput screening technology have enabled the synthesis and screening of large molecular libraries for the purposes of drug discovery. Contrary to initial expectations, the increase in screening library size, typically combined with an emphasis on compound structural diversity, did not result in a comparable increase in the number of promising hits found. In an effort to improve the likelihood of discovering hits with greater optimization potential, more recent approaches attempt to incorporate additional knowledge to the library design process to effectively guide the search. Multi-objective optimization methods capable of taking into account several chemical and biological criteria have been used to design collections of compounds satisfying simultaneously multiple pharmaceutically relevant objectives. In this chapter, we present our efforts to implement a multi-objective optimization method, MEGALib, custom-designed to the library design problem. The method exploits existing knowledge, e.g. from previous biological screening experiments, to identify and profile molecular fragments used subsequently to design compounds compromising the various objectives.

A workflow system for virtual screening in cancer chemoprevention

Computer-aided drug discovery techniques have been widely used in recent years to support the development of new pharmaceuticals. Virtual screening, the computational counterpart of experimental screening, attempts to replicate the results from in vitro and in vivo methods through the use of in silico models and tools. This paper presents the LISIs platform; a web based scientific workflow system for virtual screening that has been implemented primarily for the discovery of chemoprevention agents. We describe the overall design of the system as well as the implementation of its various components. Indicative results from early applications of the system are also presented to illustrate its potential uses and functionalities.

Open source workflow systems in life sciences informatics

A simple yet powerful programming tool enabling in silico experimentation, end-to-end data management through web services as well as use of grid and cloud processing power is scientific workflows. This technology is receiving considerable interest in recent years primarily due to its ability to promote and support scientific collaboration among large distributed research teams. The paper reviews the Scientific Workflows Management Systems (SWMS) field and investigates in detail popular open source workflow systems used commonly in life sciences informatics. Emphasis is placed on features which make these systems attractive for scientific use, e.g. user friendliness, use of distributed resources, reusability, provenance, collaboration, data integration, etc. Our conclusions indicate that although SWMS, including open source ones, have several open issues, their unique features and strong momentum clearly suggest that it is only a matter of time before they are adopted in even more scientific fields.

A Parallel implementation of a Multi-objective Evolutionary Algorithm

Multi-objective Evolutionary Algorithms (MOEAs) have features that can be exploited to harness the processing power offered by modern multi-core CPUs. Modern programming languages offer the ability to use threads and processes in order to achieve parallelism that is inherent in multi-core CPUs. In this paper we present our Parallel implementation of a MOEA algorithm and its application to the de novo drug design problem. The results indicate that using multiple processes that execute independent tasks of a MOEA, can reduce significantly the execution time required and maintain comparable solution quality thereby achieving improved performance.

Optimal graph design using a knowledge-driven multi-objective evolutionary graph algorithm

Designing appropriate graphs is a problem frequently occurring in several common applications ranging from designing communication and transportation networks to discovering new drugs. More often than not the graphs to be designed need to satisfy multiple, sometimes conflicting, objectives e.g. total length, cost, complexity or other shape and property limitations. In this paper we present our approach to solving the multi-objective graph design problem and obtaining a set of multiple equivalent compromising solutions. Our method uses multi-objective evolutionary graphs, a graph-specific meta-heuristic optimization method that combines evolutionary algorithms with graph theory and local search techniques exploiting domain-specific knowledge. In the experimental section we present results obtained for the problem of designing molecules satisfying multiple pharmaceutically relevant objectives. The results suggest that the proposed method can provide a variety of valid solutions.

Multi-objective de novo drug design using evolutionary graphs

Drug discovery and development is a complex, lengthy process and failure of a candidate molecule can occur as a result of a combination of reasons, such as poor pharmacokinetics, lack of efficacy or toxicity. Drugs compromise the numerous, sometimes competing objectives so that the benefits to patients outweigh potential drawbacks and risks [1]. De novo drug design, involves searching an immense space of feasible, drug-like molecules to select those with the highest chances of becoming drugs using computational technology [2]. Traditionally, de novo design has focused on designing molecules satisfying a single objective, such as a similarity value to a known ligand or a virtual screening score, and ignored the presence of the multiple objectives required for drug-like behavior. Recently, methods have appeared in the literature that attempt to design molecules satisfying multiple predefined objectives [3]. In this presentation we briefly review these methods and then describe a new multi-objective optimization de novo design algorithm that combines evolutionary techniques with graph-theory to directly manipulate molecular graphs and design structurally diverse molecules satisfying one or more objectives. In our experimental section we present results obtained from applying the method to design molecules with a desired biological profile with the primary constraint based on a set of known ligands. The implementation of the algorithm includes an initial step where the supplied ligand dataset is analyzed to extract and characterize frequently occurring molecular subgraphs. The resulting subgraphs together with other predefined elements form the molecular building blocks used by the algorithm. In subsequent steps a set of initial molecular graphs is prepared and subjected to an evolutionary process that involves fitness calculation against each objective, identification of a compromise surface (also known as Pareto-ranking), parent selection, mutation and crossover. Fitness calculation focuses on pharmacophoric similarity with the known ligands while parent selection uses a graph-based diversity method in order to preserve structural diversity of the evolved molecules. Our findings indicate that the proposed algorithm produces compromising solutions of substantial structural diversity and can thus be used for an efficient search of the pharmacologically interesting chemical space as defined by the supplied ligands and constrained by the objectives defined.

LiSIs: An Online Scientific Workflow System for Virtual Screening

Modern methods of drug discovery and development in recent years make a wide use of computational algorithms. These methods utilise Virtual Screening (VS), which is the computational counterpart of experimental screening. In this manner the in silico models and tools initial replace the wet lab methods saving time and resources. This paper presents the overall design and implementation of a web based scientific workflow system for virtual screening called, the Life Sciences Informatics (LiSIs) platform. The LiSIs platform consists of the following layers: the input layer covering the data file input; the pre-processing layer covering the descriptors calculation, and the docking preparation components; the processing layer covering the attribute filtering, compound similarity, substructure matching, docking prediction, predictive modelling and molecular clustering; post-processing layer covering the output reformatting and binary file merging components; output layer covering the storage component. The potential of LiSIs platform has been demonstrated through two case studies designed to illustrate the preparation of tools for the identification of promising chemical structures. The first case study involved the development of a Quantitative Structure Activity Relationship (QSAR) model on a literature dataset while the second case study implemented a docking-based virtual screening experiment. Our results show that VS workflows utilizing docking, predictive models and other in silico tools as implemented in the LiSIs platform can identify compounds in line with expert expectations. We anticipate that the deployment of LiSIs, as currently implemented and available for use, can enable drug discovery researchers to more easily use state of the art computational techniques in their search for promising chemical compounds. The LiSIs platform is freely accessible (i) under the GRANATUM platform at: http://www.granatum.org and (ii) directly at: http://lisis.cs.ucy.ac.cy.

Deriving Quantitative Structure-Activity Relationship Models Using Genetic Programming for Drug Discovery

Genetic Programming is a heuristic search algorithm inspired by evolutionary techniques that has been shown to produce satisfactory solutions to problems related to several scientific domains [1]. Presented here is a methodology for the creation of Quantitative Structure-Activity Relationship (QSAR) models for the prediction of chemical activity, using Genetic Programming. QSAR analysis is crucial for drug discovery since good QSAR models enable human experts to select compounds with increased chances of being active for further investigations. Our technique has been tested using the Selwood dataset, a benchmark dataset for the QSAR field [2]. The results indicate that the QSAR models created are accurate, reliable and simple and can thus be used to identify molecular descriptors correlated with measured activity and for the prediction of the activity of untested molecules. The QSAR models we generated predict the activity of untested molecules with an error ranging between 0.46 -0.8 on the scale [-1,1]. These results compare favourably with results sited in the literature for the same dataset [3], [4], Our models are constructed using any combination of the arithmetic operators {+, -, /, *}, the descriptors available and constant values.

Knowledge-driven multi-objective de novo drug design

Drug discovery is an inherently multi-objective process since drugs need to satisfy not only activity requirements but also a range of other properties such as selectivity and toxicity. However, drug discovery process practices, including both experimental and computational methods, commonly ignore this fact and focus on a single pharmaceutical objective at a time. De novo design, the branch of chemoinformatics addressing the in silico design of ligands from scratch, follows a similar approach typically focusing on a single objective, such as an interaction score to a target receptor or similarity to a known drug [1]. Recently, methods have appeared in the literature that attempt to design molecules satisfying multiple predefined objectives [2]. Motivated from the initial success of these algorithms [3] – as well as their widespread use in other scientific fields – we have preciously introduced MEGA, a Multi-objective Evolutionary Graph Algorithm with the aim of performing de novo design taking into account numerous pharmaceutically relevant objectives. Unlike most other evolutionary-based de novo algorithms, MEGA uses graph data structures for chromosome representation and directly manipulates the graphs to perform a global search for promising solutions. The initial version of the algorithm includes problem-domain specific knowledge in the form of weighted molecular fragments used during chemical structure evolution. Capitalizing on lessons learned we have designed an extension blending additional problem knowledge and local search capabilities to achieve faster convergence. This type of algorithm, commonly referred to as Memetic in the optimization community, has been shown to be orders of magnitude faster than traditional evolutionary algorithms [4] especially in problems searching large, complex and multimodal solution surfaces.