Ligand docking is a widely used approach in virtual screening. In recent years a large number of publications have appeared in which docking tools are compared and evaluated for their effectiveness in virtual screening against a wide variety of protein targets. These studies have shown that the effectiveness of docking in virtual screening is highly variable due to a large number of possible confounding factors. Another class of method that has shown promise in virtual screening is the shape-based, ligand-centric approach. Several direct comparisons of docking with the shape-based tool ROCS have been conducted using data sets from some of these recent docking publications. The results show that a shape-based, ligand-centric approach is more consistent than, and often superior to, the protein-centric approach taken by docking.

Comparison of Shape-Matching and Docking as Virtual Screening Tools

This paper tries to explore the optimum variable selection strategy for Partial Least Squares (PLS) regression using a model dataset of cytoprotection data. The compounds of the dataset were classified using K-means clustering technique applied on standardized descriptor matrix and ten combinations of training and test sets were generated based on the obtained clusters. For a particular training set, PLS models were developed with a number of components optimized by leave-one-out Q2 and then the developed models were validated (externally) using the test set compounds. For each set, PLS model was initially constructed using all descriptors (variables). The variables having least standardized values of regression coefficients were deleted and the next model was developed with a reduced set of variables. These steps were performed several times until further reduction in number of variables did not improve Q2 value. In each case, statistical parameters like predictive R2 (R2pred), squared correlation coefficient between observed and predicted values with (r2) and without () intercept and Root Mean Square Error of Prediction (RMSEP) were calculated from the test set compounds. In case of all ten sets, Q2 values steadily increase on deletion of variables while R2pred values do not show any specific trend. In no case, the highest Q2 and highest R2pred appear in the same trial, i.e., with the same combinations of variables. This suggests that from the viewpoint of external predictability, choice of variables for PLS based on Q2 value may not be optimum. Moreover, a clear separation of r2 and r02 curves in some sets suggests that such models may not be truly predictive in spite of acceptable R2pred values. Another observation is that coefficient of determination R2 for the training set is more immune to changes on deletion of variables than the validation parameters like Q2 and R2pred. Finally, a new parameter rm2 has been suggested to indicate external predictability of QSAR models.

On Some Aspects of Variable Selection for Partial Least Squares Regression Models

Machine learning in chemoinformatics and drug discovery.

Prediction of chemical bioactivity and physical properties has been one of the most important applications of statistical and more recently, machine learning and artificial intelligence methods in chemical sciences. This field of research, broadly known as quantitative structure–activity relationships (QSAR) modeling, has developed many important algorithms and has found a broad range of applications in physical organic and medicinal chemistry in the past 55+ years. This Perspective summarizes recent technological advances in QSAR modeling but it also highlights the applicability of algorithms, modeling methods, and validation practices developed in QSAR to a wide range of research areas outside of traditional QSAR boundaries including synthesis planning, nanotechnology, materials science, biomaterials, and clinical informatics. As modern research methods generate rapidly increasing amounts of data, the knowledge of robust data-driven modelling methods professed within the QSAR field can become essential for scientists working both within and outside of chemical research. We hope that this contribution highlighting the generalizable components of QSAR modeling will serve to address this challenge.

QSAR without borders

The identification of interactions between drugs/compounds and their targets is crucial for the development of new drugs. In vitro screening experiments (i.e. bioassays) are frequently used for this purpose; however, experimental approaches are insufficient to explore novel drug-target interactions, mainly because of feasibility problems, as they are labour intensive, costly and time consuming. A computational field known as 'virtual screening' (VS) has emerged in the past decades to aid experimental drug discovery studies by statistically estimating unknown bio-interactions between compounds and biological targets. These methods use the physico-chemical and structural properties of compounds and/or target proteins along with the experimentally verified bio-interaction information to generate predictive models. Lately, sophisticated machine learning techniques are applied in VS to elevate the predictive performance. The objective of this study is to examine and discuss the recent applications of machine learning techniques in VS, including deep learning, which became highly popular after giving rise to epochal developments in the fields of computer vision and natural language processing. The past 3 years have witnessed an unprecedented amount of research studies considering the application of deep learning in biomedicine, including computational drug discovery. In this review, we first describe the main instruments of VS methods, including compound and protein features (i.e. representations and descriptors), frequently used libraries and toolkits for VS, bioactivity databases and gold-standard data sets for system training and benchmarking. We subsequently review recent VS studies with a strong emphasis on deep learning applications. Finally, we discuss the present state of the field, including the current challenges and suggest future directions. We believe that this survey will provide insight to the researchers working in the field of computational drug discovery in terms of comprehending and developing novel bio-prediction methods.

/pdf/recent-applications-of-deep-learning-and-machine-2jhuk5h77l.pdf

Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases.

Hierarchical clustering algorithms such as Wards or complete-link are commonly used in compound selection and diversity analysis. Many such applications utilize binary representations of chemical structures, such as MACCS keys or Daylight fingerprints, and dissimilarity measures, such as the Euclidean or the Soergel measure. However, hierarchical clustering algorithms can generate ambiguous results owing to what is known in the cluster analysis literature as the ties in proximity problem, i.e., compounds or clusters of compounds that are equidistant from a compound or cluster in a given collection. Ambiguous ties can occur when clustering only a few hundred compounds, and the larger the number of compounds to be clustered, the greater the chance for significant ambiguity. Namely, as the number of "ties in proximity" increases relative to the total number of proximities, the possibility of ambiguity also increases. To ensure that there are no ambiguous ties, we show by a probabilistic argument that the number of compounds needs to be less than 2(n 1/4), where n is the total number of proximities, and the measure used to generate the proximities creates a uniform distribution without statistically preferred values. The common measures do not produce uniformly distributed proximities, but rather statistically preferred values that tend to increase the number of ties in proximity. Hence, the number of possible proximities and the distribution of statistically preferred values of a similarity measure, given a bit vector representation of a specific length, are directly related to the number of ties in proximities for a given data set. We explore the ties in proximity problem, using a number of chemical collections with varying degrees of diversity, given several common similarity measures and clustering algorithms. Our results are consistent with our probabilistic argument and show that this problem is significant for relatively small compound sets.

Ties in proximity and clustering compounds.

We have found that molecular shape and electrostatics, in conjunction with 2D structural fingerprints, are important variables in discriminating classes of active and inactive compounds. The subject of this paper is how to explore the selection of these variables and identify their relative importance in quantitative structure-activity relationships (QSAR) analysis. We show the use of these variables in a form of similarity searching with respect to a crystal structure of a known bound ligand. This analysis is then validated through k-fold cross-validation of enrichments via several common classifiers. Additionally, we show an effective methodology using the variables in hypothesis generation; namely, when the crystal structure of a bound ligand is not known.

Variable selection and model validation of 2D and 3D molecular descriptors

Chemoinformatics applications of cluster analysis over the past 35 years include chemical diversity for compound acquisition, analysis of HTS results for lead discovery, 2D and 3D chemical similarity searching for virtual screening, and hypothesis generation for lead hopping using molecular shape and pharmacophore descriptors. These applications still provide the majority of cluster analysis usage, but the advent of greater and greater computational resources has allowed researchers to tackle applications of ever increasing scale and complexity. In the past few years, a far broader array of clustering methods is now used—some entirely new, some common to other disciplines, and others modified to specific chemoinformatic applications. The chemoinformatic applications have also broadened to include greater biological information more commonly associated with bioinformatics. Indeed, clustering techniques, such as coclustering or self-organizing trees, commonly found in bioinformatics, are beginning to find chemoinformatic application uses. Issues such as visualization and validation of clustering results continue to present challenging problems, especially given that the scale of many problems now attempted has increased enormously. Some new validation techniques have been introduced in the chemoinformatics literature that now allow for both a better understanding of the clustering results and help point to methods of greater efficacy. Effective validation and visualization of clustering results of large data sets has proven to be more problematic. WIREs Comput Mol Sci 2014, 4:34–48. doi: 10.1002/wcms.1152



The authors have declared no conflicts of interest in relation to this article.



For further resources related to this article, please visit the WIREs website.

Chemoinformatics applications of cluster analysis

In this paper we present a user interface, CANDID Camera, for image retrieval using query-by-example technology. Included in the interface are several new layout algorithms based on multidimensional scaling techniques that visually display global and local relationships between images within a large image database. We use the CANDID project algorithms to create signatures of the images, and then measure the dissimilarity between the signatures. The layout algorithms are of two types. The first are those that project the all-pairs dissimilarities to two dimensions, presenting a many-to-many relationship for a global view of the entire database. The second are those that relate a query image to a small set of matched images for a one-to-many relationship that provides a local inspection of the image relationships. Both types are based on well-known multidimensional scaling techniques that have been modified and used together for efficiency and effectiveness. They include nonlinear projection and classical projection. The global maps are hybrid algorithms using classical projection together with nonlinear projection. We have developed several one-to-many layouts based on a radial layout, also using modified nonlinear and classical projection.

/pdf/interactive-layout-mechanisms-for-image-database-retrieval-4hzbntx7qe.pdf

Interactive layout mechanisms for image database retrieval

Introduction History Bioinformatics and Drug Discovery Statistical Learning Theory and Exploratory Data Analysis Clustering Algorithms Computational Complexity Data Types Normalization and Scaling Transformations Formats Data Matrices Measures of Similarity Proximity Matrices Symmetric Matrices Dimensionality, Components, Discriminants Graph Theory Clustering Forms Partitional Hierarchical Mixture Models Sampling Overlapping Fuzzy Self-Organizing Hybrids Partitional Algorithms K-Means Jarvis-Patrick Spectral Clustering Self-Organizing Maps Cluster Sampling Algorithms Leader Algorithms Taylor-Butina Algorithm Hierarchical Algorithms Agglomerative Divisive Hybrid Algorithms Self-Organizing Tree Algorithm Divisive Hierarchical K-Means Exclusion Region Hierarchies Biclustering Asymmetry Measures Algorithms Ambiguity Discrete Valued Data Types Precision Ties in Proximity Measure Probability and Distributions Algorithm Decision Ambiguity Overlapping Clustering Algorithms Based on Ambiguity Validation Validation Measures Visualization Example Large Scale and Parallel Algorithms Leader and Leader-Follower Algorithms Taylor-Butina K-Means and Variants Examples Appendices Bibliography A Glossary and Exercises appear at the end of each chapter.

John MacCuish

Papers

Ties in proximity and clustering compounds.

Variable selection and model validation of 2D and 3D molecular descriptors

Chemoinformatics applications of cluster analysis

Interactive layout mechanisms for image database retrieval

Clustering in Bioinformatics and Drug Discovery