Classification and recognition of compounds in low-resolution open-path FT-IR spectrometry by Kohonen self-organizing maps

The possibility of using one- and two-dimensional Kohonen self-organizing maps (SOMs) to recognize similarities in low-resolution vapor-phase infrared spectra without any additional information, i.e., in an unsupervised mode, has been investigated. Full-range vapor-phase FT-IR reference spectra were first used to train the networks and the trained networks were then used to classify the reference spectra into several groups. The feasibility of reducing the spectral range to be consistent with the atmospheric windows used in open-path FT-IR spectrometry was also studied. Kohonen networks are shown to be relatively immune to the presence of noise. An example of using a trained Kohonen map to recognize the presence of selected compounds in field-measured open-path FT-IR spectra is given.     

Classification and recognition of compounds in low-resolution open-path FT-IR spectrometry by Kohonen self-organizing maps

The possibility of using one- and two-dimensional Kohonen self-organizing maps (SOMs) to recognize similarities in low-resolution vapor-phase infrared spectra without any additional information, i.e., in an unsupervised mode, has been investigated. Full-range vapor-phase FT-IR reference spectra were first used to train the networks and the trained networks were then used to classify the reference spectra into several groups. The feasibility of reducing the spectral range to be consistent with the atmospheric windows used in open-path FT-IR spectrometry was also studied. Kohonen networks are shown to be relatively immune to the presence of noise. An example of using a trained Kohonen map to recognize the presence of selected compounds in field-measured open-path FT-IR spectra is given. Received: 13 January 1998 / Revised: 4 May 1998 / Accepted: 14 May 1998     

Ligand-based combinatorial design of selective purinergic receptor (A2A) antagonists using self-organizing maps

A virtual screening procedure based on a topological pharmacophore similarity metric and self-organizing maps (SOM) was developed and applied to optimizing combinatorial products functioning as P(1) purinergic receptor antagonists. The target was the human A(2A) receptor. A SOM was developed using a set of biologically tested molecules to establish a preliminary structure-activity relationship. A combinatorial library design was performed by projecting virtually assembled new molecules onto the SOM. A small focused library of 17 selected combinatorial products was synthesized and tested. On average, the designed structures yielded a 3-fold smaller binding constant ( approximately 33 vs approximately 100 nM) and 3.5-fold higher selectivity (50 vs 14) than the initial library. The most selective compound obtained revealed a 121-fold relative selectivity for A(2A) with K(i) (A(2A)) = 2.4 nM, and K(i) (A(1)) = 292 nM. This result demonstrates that it was possible to design a small, activity-enriched focused library with an improved property profile using the SOM virtual screening approach. The strategy might be particularly useful in projects in which structure-based design cannot be applied because of a lack of receptor structure information, for example, in the many projects aiming at finding new GPCR modulators.     

Spectral pattern recognition using self-organizing MAPS

A Kohonen neural network is an iterative technique used to map multivariate data. The network is able to learn and display the topology of the data. Self-organizing maps have advantages as well as drawbacks when compared to principal component plots. One advantage is that data preprocessing is usually minimal. Another is that an outlier will only affect one map unit and its neighborhood. However, outliers can have a drastic and disproportionate effect on principal component plots. Removing them does not always solve the problem for as soon as the worst outliers are deleted, other data points may appear in this role. The advantage of using self-organizing maps for spectral pattern recognition is demonstrated by way of two studies recently completed in our laboratory. In the first study, Raman spectroscopy and self-organizing maps were used to differentiate six common household plastics by type for recycling purposes. The second study involves the development of a potential method to differentiate acceptable lots from unacceptable lots of avicel using diffuse reflectance near-infrared spectroscopy and self-organizing maps.     

Computational methods and software in computer-aided combinatorial library design

Recent trends in the computer-aided design of diverse and focussed combinatorial libraries are surveyed. First, chemical data input, storage and retrieval including chemical database management and virtual chemical structure enumeration are outlined as background. Then, the optimization of ADMET parameters, diversity maximization, molecular similarity search, QSAR-based virtual screening, pharmacophore search and molecular docking are discussed.     

CombiDOCK: Structure-based combinatorial docking and library design

We have developed a strategy for efficiently docking a large combinatorial library into a target receptor. For each scaffold orientation, all potential fragments are attached to the scaffold, their interactions with the receptor are individually scored and factorial combinations of fragments are constructed. To test its effectiveness, this approach is compared to two simple control algorithms. Our method is more efficient than the controls at selecting best scoring molecules and at selecting fragments for the construction of an exhaustive combinatorial library. We also carried out a retrospective analysis of the experimental results of a 10×10×10 exhaustive combinatorial library. An enrichment factor of approximately 4 was found for identifying the compounds in the library that are active at 330 nM.     

A novel approach to combinatorial library design

We address the problem of designing a general-purpose combinatorial library to screen for pharmaceutical leads. Conventional approaches focus on diversity as the primary factor in designing such libraries. We suggest making screening libraries out of a set of pharmaceutically relevant scaffolds, with multiple analogs per scaffold. The rationale for this rests on the fact that even though the hit-rate in active series is much higher than in the database as a whole, often a large fraction of the compounds in active series are inactive. This is especially true when the series has not been optimized for the target under study. We introduce the concept of hit-rate within a series and use historic screening data to arrive at a crude estimate for it. We then use simple probability arguments to show that 50-100 compounds are required in each series in order to be nearly certain of finding at least one active compound in each true active series for any given target.     

Automated analysis of proton NMR spectra from combinatorial rapid parallel synthesis using self-organizing maps

It is now quite routine to acquire proton NMR spectra of compounds in 96-well plates prepared in a rapid parallel synthesis fashion using a flow-NMR automation setup. However, the analysis of 96 NMR spectra obtained in this manner is often laborious and painstakingly slow. We have developed a new, automated method for rapidly analyzing 96 NMR spectra of compounds synthesized in an 8 x 12 matrix using self-organizing maps (SOM). This unsupervised neural network is capable of clustering together NMR spectra containing a common pattern of -R groups and identifying outliers from within such clusters. Analysis of these outlier spectra can quickly help indicate the presence of undesired products, impurities, starting materials, and other unexpected errors in a 96-well plate synthesis by focusing the chemists' attention on the aberrant NMR spectra. Thus, SOM can be a valuable tool in performing efficient quality control on combinatorial libraries.     

Application of self-organizing maps in conformational analysis of lipids.

The characteristics of lipid assemblies are important for the functions of biological membranes. This has led to an increasing utilization of molecular dynamics simulations for the elucidation of the structural features of biomembranes. We have applied the self-organizing map (SOM) to the analysis of the complex conformational data from a 1-ns molecular dynamics simulation of PLPC phospholipids in a membrane assembly. Mapping of 1.44 million molecular conformations to a two-dimensional array of neurons revealed, without human intervention, the main conformational features in hours. Both the whole molecule and the characteristics of the unsaturated fatty acid chains were analyzed. All major structural features were easily distinguished, such as the orientational variability of the headgroup, the mainly trans state dihedral angles of the sn-1 chain, and both straight and bent conformations of the unsaturated sn-2 chain. Furthermore, presentation of the trajectory of an individual lipid molecule on the map provides information on conformational dynamics. The present results suggest that the SOM method provides a powerful tool for routinely gaining rapid insight to the main molecular conformations as well as to the conformational dynamics of any simulated molecular assembly without the requirement of a priori knowledge.     

Rational principles of compound selection for combinatorial library design

It is practically impossible in a short period of time to synthesize and test all compounds in any large exhaustive chemical library. We discuss rational approaches to selecting representative subsets of virtual libraries that help direct experimental synthetic efforts for both targeted and diverse library design. For targeted library design, we consider principles based on the similarity to lead molecules. In the case of diverse library design, we discuss algorithms aimed at the selection of both diverse and representative subsets of the entire chemical library space. We illustrate methodologies with several practical examples.     

Sensitivity analysis and other improvements to tailored combinatorial library design

"Tailoring" combinatorial libraries was developed several years ago as a very general and intuitive method to design diverse compound collections while controlling the profile of other pharmaceutically relevant properties. The candidate substituents were assigned to "categorical bins" according to their properties, and successive steps of D-optimal design were performed to generate diverse substituent sets consistent with required membership quotas from each bin. This serial algorithm was expedient to implement from existing D-optimal design codes, but was order-dependent and did not generally locate the very best possible design. A new "parallel" Fedorov search algorithm has now been implemented that can find the most diverse property-tailored design. An ambiguous mass penalty has been added, whereby most duplicate masses can be eliminated with little loss of library diversity. Sensitivity analysis has also been added to quantitatively explore the diversity trade-offs due to increasing or decreasing each specific kind of bias.     

Solid phase combinatorial synthesis of a library of macro-heterocycles and related acyclic compounds

A solid phase synthesis of macrolactones from three building blocks and in eight steps is described. The synthesis which is carried out on the DHP resin includes Mitsunobu and DIC couplings. The macrocyclization occurs by SN2 displacement of an allylic chloride by a malonate anion. The synthetic methodology is suitable for the synthesis of arrays of macrocycles as well as linear compounds.     

A scalable approach to combinatorial library design for drug discovery

In this paper, we propose an algorithm for the design of lead generation libraries required in combinatorial drug discovery. This algorithm addresses simultaneously the two key criteria of diversity and representativeness of compounds in the resulting library and is computationally efficient when applied to a large class of lead generation design problems. At the same time, additional constraints on experimental resources are also incorporated in the framework presented in this paper. A computationally efficient scalable algorithm is developed, where the ability of the deterministic annealing algorithm to identify clusters is exploited to truncate computations over the entire data set to computations over individual clusters. An analysis of this algorithm quantifies the tradeoff between the error due to truncation and computational effort. Results applied on test data sets corroborate the analysis and show improvement by factors as large as 10 or more, depending on the data sets.     

OptDesign: extending optimizable k-dissimilarity selection to combinatorial library design

Optimizable k-dissimilarity (OptiSim) selection entails drawing a series of subsamples of size k from a population and choosing the "best" candidate from each such subsample for inclusion in the selection set. By varying the size of the subsample, one can control the balance between representativeness and diversity in the selection set obtained. In the original formulation, a uniform random sampling from among valid candidates was used to draw the subsamples from a single target population. Here we describe in detail two key modifications that serve to extend the OptiSim methodology to vector selection for interdependent variables, specifically as applied to the design of combinatorial sublibraries. The first modification involves pivoting between variables: subsamples are drawn from each reagent pool in turn, with the viability of each candidate being evaluated in isolation as well as in terms of the products it will produce from complementary reagents already selected. The filters applied may be static or dynamic in nature, with molecular weight and hydrophobicity being examples of the former and structural diversity with respect to reagents already selected being an example of the latter. The second key modification is adding the ability to bias the selection of candidate reagents for inclusion in the subsamples. Taken together, these modifications support the efficient generation of multiblock and other sparse matrix designs that are both representative and diverse, and for which "backfilling" of designs edited to remove undesirable reagents or products is straightforward. The method is intrinsically fast and efficient, since enumeration of the full combinatorial is not required- only those candidates actually considered for inclusion need be evaluated. Moreover, because the subsample selection step is separate from the diversity-based selection of the "best" candidate, incorporating such bias in favor of a competing criterion such as low price provides a "natural," nonparametric mechanism for generating designs that are likely to be "good" in a double-objective, Pareto sense.     

Efficient solution phase combinatorial access to a library of pyrazole- and triazole-containing compounds

An efficient method for the synthesis of multidentate molecules, with various building blocks, such as pyrazolyl, triazolyl, pyridyl, furyl, or bispinidyl derivatives, is reported. This approach is based on the condensation of easily available N-alkyl heteroarylamines with N-hydroxymethyl pyrazoles or triazoles.     

Scaffold diversity of natural products: inspiration for combinatorial library design

Natural products contain scaffold structures that can be systematically exploited for the design of combinatorial compound libraries with druglike properties. We review approaches for scaffold identification, and compare properties and pharmacophoric features of drugs and natural products. In particular, an application of the self-organizing map technique is presented for natural product-derived compound and library design.     

Application of pattern recognition and piezoelectric sensor array for the detection of organic compounds

Organic vapours were measured by an array of piezoelectric crystal detectors. Quartz crystals were coated by different GC stationary phases. Four coated crystals were placed in an array and pattern recognition was used for identification of the compounds including acetone, benzene, chloroform and pentane. A computer program was developed for the measurement of the frequency changes and data processing. Pattern recognition method using feature extraction was applied for identification of analytes.