Problems in data retrieval systems for analytical spectroscopy

Some of the existing retrieval systems for analytical purposes based on different spectrometric data are described and compared with systems under development. Requirements concerning input data, coding of fragments or complete chemical structures, preprocessing of input data, and evaluation of the retrieved results are discussed, and some examples are given. Some trends for future developments in this field are outlined.     

An example of 2-block predictive partial least-squares regression with simulated data

Practical examples with simulated data are described to illustrate the material described in the preceding tutorial on partial least-squares regression. Starting from the perfect model for two dimensions, noise, nonlinearities and interference are added gradually in order to study their influence. The examples are related to calibration in chemical analysis.     

Numerical data acquisition,dissemination and retrieval systems

Scientific data acquisition, dissemination and retrieval methods are continuing to develop. As both information specialists and working scientists realize the need for, and value of, good scientific data, there is more activity in this area. Some problems in this field are described. The creation of a knowledge base for spectroscopy is proposed as a means of helping to solve the problem of chemical structure elucidation.     

Aerial robotic data acquisition system

A small, unmanned aerial vehicle (UAV) equipped with sensors for physical and chemical measurements of remote environments, is described. A miniature helicopter airframe is used as a platform for sensor testing and development. The sensor output is integrated with the flight control system for real-time, interactive, data acquisition and analysis. Pre-programmed flight missions will be flown with several sensors to demonstrate the cost-effective surveillance capabilities of this new technology.     

From data point timelines to a well curated data set, data mining of experimental data and chemical structure data from scientific articles, problems and possible solutions

The scientific literature is important source of experimental and chemical structure data. Very often this data has been harvested into smaller or bigger data collections leaving the data quality and curation issues on shoulders of users. The current research presents a systematic and reproducible workflow for collecting series of data points from scientific literature and assembling a database that is suitable for the purposes of high quality modelling and decision support. The quality assurance aspect of the workflow is concerned with the curation of both chemical structures and associated toxicity values at (1) single data point level and (2) collection of data points level. The assembly of a database employs a novel “timeline” approach. The workflow is implemented as a software solution and its applicability is demonstrated on the example of the Tetrahymena pyriformis acute aquatic toxicity endpoint. A literature collection of 86 primary publications for T. pyriformis was found to contain 2,072 chemical compounds and 2,498 unique toxicity values, which divide into 2,440 numerical and 58 textual values. Every chemical compound was assigned to a preferred toxicity value. Examples for most common chemical and toxicological data curation scenarios are discussed.     

A co‐training algorithm for multi‐view data with applications in data fusion

In several scientific applications, data are generated from two or more diverse sources (views) with the goal of predicting an outcome of interest. Often it is the case that the outcome is not associated with any single view. However, the synergy of all measurements from each view may yield a more predictive classifier. For example, consider a drug discovery application in which individual molecules are described partially by several assay screens based on diverse profiles and partially by their chemical structural fingerprints. A common classification problem is to determine whether the molecule is associated with a particular disease. In this paper, a co‐training algorithm is developed to utilize data from diverse sources to predict the common class variable. Novel enhancements for variable importance, robustness to a mislabeled class variable, and a technique to handle unbalanced classes are applied to the motivating data set, highlighting that the approach attains strong performance and provides useful diagnostics for data analytic purposes. In addition, comparisons to a framework with data fusion using partial least squares (PLS) are also assessed on real data. An R package for performing the proposed approach is provided as Supporting information. Copyright © 2003 John Wiley & Sons, Ltd.     

Chemical data: an essential tool in the regulation of drugs.

How the Bureau of Drugs laboratories and offices obtain chemical data from the scientific literature, from user complaints and product defect reporting systems, from the drug manufacturers, from analyses of drug samples collected from the market, and from analytical research are described. The chemical data thus educed have been used successfully in developing new analytical methods, in establishing better specifications of drug quality, in removing adulterated drugs from the marketplace, in successfully prosecuting purveyors of substandard drugs, and in general assuring that consumers are provided with safe and effective drugs of high quality.     

ThermoData Engine (TDE): software implementation of the dynamic data evaluation concept. 5. Experiment planning and product design

ThermoData Engine (TDE) is the first full-scale software implementation of the dynamic data evaluation concept, as reported recently in this journal. In the present paper, we describe development of an algorithmic approach to assist experiment planning through assessment of the existing body of knowledge, including availability of experimental thermophysical property data, variable ranges studied, associated uncertainties, state of prediction methods, and parameters for deployment of prediction methods and how these parameters can be obtained using targeted measurements, etc., and, indeed, how the intended measurement may address the underlying scientific or engineering problem under consideration. A second new feature described here is the application of the software capabilities for aid in the design of chemical products through identification of chemical systems possessing desired values of thermophysical properties within defined ranges of tolerance. The algorithms and their software implementation to achieve this are described. Finally, implementation of a new data validation and weighting system is described for vapor-liquid equilibrium (VLE) data, and directions for future enhancements are outlined.     

A combined linear and nonlinear factor analysis program package for chemical data evaluation

An underlying variable factor analysis program (UVFA) is described. The theories of principal component analysis and nonlinear least-squares projection techniques are outlined and compared. Several applications from various chemical fields are presented which show that a complete analysis of the underlying structure and dimensionality of a chemical data set should always include these nonlinear projection techniques.     

Using animal dosimetry models to interpret human bioassay data for actinide exposures

Bioassay and dosometry models are needed to estimate intakes of radionuclides, and to calculate radiation doses to target tissues following such in takes. Because of the diversity of exposure materials, individual biological variabilities, and the general lack of adequate bioassay information and knowledge of the metabolism of radionuclides, current models are based mostly on empiricism. This paper describes biokinetic/dosimetry models for U, Am, and Cm. They are based on experimental data developed from studies in dogs that inhaled one of the above radionuclides in specific chemical forms and specific particle sizes. The models, which are based on similar biological principles, and, therefore, have similar structure, are applied to the very sparse human bioassay data available from cases of exposure to either U, Am, or Cm. The results thus far indicate that the lung retention for the different actinides are well described by the models, that urinary bioassay data can be described within limited time periods, and that the fecal excretion rate is not adequately described. Improvements in modeling are predicted on increased publication of human bioassay data, and better cooperative interaction between model developers and health protection professionals responsible for industrial bioassay programs.     

Estimation of kinetic parameters from thermochemical data

A method is described by which kinetic parameters may be calculated from the measured temperature changes caused by the heat produced during a chemical reaction. An isoperibol titration calorimeter with an ampoule-breaking facility is used to obtain the temperature data. The temperature changes resulting from the reaction between tri-isopropyl phosphite and sulphur (S(8)) are used as an example to demonstrate the effectiveness of the method. The temperature changes are used to calculate an enthalpy of reaction. From the enthalpy of reaction and intermediate heats, instantaneous concentrations of the reactants may be calculated.     

New global communication process in thermodynamics: impact on quality of published experimental data

Thermodynamic data are a key resource in the search for new relationships between properties of chemical systems that constitutes the basis of the scientific discovery process. In addition, thermodynamic information is critical for development and improvement of all chemical process technologies. Historically, peer-reviewed journals are the major source of this information obtained by experimental measurement or prediction. Technological advances in measurement science have propelled enormous growth in the scale of published thermodynamic data (almost doubling every 10 years). This expansion has created new challenges in data validation at all stages of the data delivery process. Despite the peer-review process, problems in data validation have led, in many instances, to publication of data that are grossly erroneous and, at times, inconsistent with the fundamental laws of nature. This article describes a new global data communication process in thermodynamics and its impact in addressing these challenges as well as in streamlining the delivery of the thermodynamic data from "data producers" to "data users". We believe that the prolific growth of scientific data in numerous and diverse fields outside thermodynamics, together with the demonstrated effectiveness and versatility of the process described in this article, will foster development of such processes in other scientific fields.     

Spectra–structure relationships in carbon-13 nuclear magnetic resonance spectroscopy. Results from a large data base

A link between a substructure searching system and a ¹³C NMR data base has been established and permits the retrieval and examination of the chemical shifts associated with specific substructures. The means by which these searches are accomplished is described and the results from the searches are presented and discussed. The system is interactive, and can be used to locate in the data base the chemical shifts of carbon atoms in precisely defined environments. Alternatively, it may be used to learn the range of the chemical shifts possessed by particular types of carbon atoms, such as N-methyl or O-methyl carbons.     

Disclose: an integrated set of multivariate display procedures for chemical and pharmaceutical data

An approach to providing thorough, reliable views of multivariate data is described using an integrated set of classification and display routines (DISCLOSE). The programs are mainly based on standard, published algorithms. The techniques include principal components plots, non-linear mapping, three hirarchal cluster analysis algorithms, minimum spanning tree, amd nearest-neighbour representation, together with alternative methods for variable normalization. DISCLOSE, which does not require computer graphics hardware, is well suited for providing a systematic initial visualization of data-set structure, and is applicable to chemical, pharmaceutical, and biological data.     

Sequential superparamagnetic clustering for unbiased classification of high-dimensional chemical data

For the clustering of chemical structures that are described by the Similog, ISIS count, and ISIS binary fingerprints, we propose a sequential superparamagnetic clustering approach. To appropriately handle nonbinary feature keys, we introduce an extension of the binary Tanimoto similarity measure. In our applications, data sets composed of structures from seven chemically distinct compound classes are evaluated and correctly clustered. The comparison, with results from leading methods, indicates the superiority of our sequential superparamagnetic clustering approach.     

The use of glass electrodes for the determination of formation constants-III Optimization of titration data: The esta library of computer programs

Difficulties with the analysis of titration data for the determination of formation constants are discussed. Considerable differences between published values can, in part, be attributed to incorrect "model selection", i.e., incorrect selection of chemical species. Experimental errors also contribute to the problem in ways that are neither well understood nor easy to overcome. The library of computer programs described, called ESTA (Equilibrium Simulation for Titration Analysis), has been written to investigate better methods of optimizing formation constants from potentiometric titration data.     

A computer system for structural identification of organic compounds from 13C NMR data

A computerised library search system for ¹³C NMR data is described. Given the spectral data of an unknown compound, the system will retrieve from the library those reference compounds exhibiting similar spectra. For comparison the spectral data are converted into a binary code, designed to reflect the underlying structure rather than exact values for chemical shifts. Thus, the ability of the system to retrieve compounds similar to the unknown (as opposed to identical) is greatly enhanced. A sophisticated search strategy adapting itself automatically to the problem at hand makes the system highly efficient.     

Multivariate analysis of time-resolved mass spectral data

Multvariate analysis of time-resolved pyrolysis/mass spectrometric data is described. The approach is based on the variance diagram (VARDIA), a recently developed technique that quantifies the clustering of variables in two-dimensional factor analysis (sub)-spaces in a rotational scanning procedure. A maximum in the VARDIA plot indicates a correlated behavior of the mass variables, indicating a common origin. This common origin is generally caused by a change in the concentration of a chemical component. With this information the “factor spectrum” and the scores of the component can be retrieved. For time-resolved serial data, consideration of the clustering behavior of the variables as a function of time is more appropriate than a rotational scanning procedure. Adaptation of the VARDIA for serial data, such as time-resolved data, is described. This approach has the advantage that all the factors can be used. It will be shown that the resolution of the obtained curve than the total ion current curve as a function of time. Examples will be given for time-resolved data of coal, rubber and wood samples.