Computer-Enhanced Analytical Spectroscopy MODERN ANALYTICAL CHEMISTRY Series Editor: David Hercules University of Pittsburgh ANALYTICAL ATOMIC SPECTROSCOPY William G. Schrenk APPLIED ATOMIC SPECTROSCOPY Volumes 1 and 2 Edited by E. L. Grove CHEMICAL DERIVA TIZAT ION IN ANALYTICAL CHEMISTRY Edited by R. W. Frei and J. F. Lawrence Volume 1: Chromatography Volume 2: Separation and Continuous Flow Techniques COMPUTER-ENHANCED ANALYTICAL SPECTROSCOPY Edited by Henk L. C. Meuzelaar and Thomas L. Isenhour ION-SELECTIVE ELECTRODES IN ANALYTICAL CHEMISTRY Volumes 1 and 2 Edited by Henry Freiser MODERN FLUORESCENCE SPECTROSCOPY Volumes 1-4 Edited by E. L. Wehry PHOTOELECTRON AND AUGER SPECTROSCOPY Thomas A. Carlson TRANSFORM TECHNIQUES IN CHEMISTRY Edited by Peter R. Griffiths Computer-Enhanced Analytical Spectroscopy Edited by Henk L. C. Meuzelaar The University oj Utah Salt Lake City, Utah and Thomas L. Isenhour Utah State University Logan, Utah Plenum Press • New York and London Library of Congress Cataloging in Publication Data Computer-enhanced analytical spectroscopy / edited by Henk L. C. Meuzelaar and Thomas L. Isenhour. p. cm. - (Modern analytical chemistry) Papers from the First Hidden Peak Symposium, held at Snowbird, Utah, June 1986. Includes bibliographies and index. Contents: Optimization and exploratory data analysis. Development of an AI-based optimization system for tandem mass spectrometry / Carla M. Wong and Hal R. Brand. Curve-fitting and Fourier self-deconvolution for the quantitative representation of complex spectra / Peter R. Griffiths, John A. Pierce, and Gao Hongjin ... [etc] ISBN-13: 978-1-4684-5370-6 e-ISBN-13: 978-1-4684-5368-3 DOl: 10.1007/978-1-4684-5368-3 1. Spectrum analysis-Data processing. 1. Meuzelaar, Henk L. C. II. Isenhour, Thomas L. III. Hidden Peak Symposium (1st: 1986: Snowbird, Utah) IV. Series. QD95.C6323 1987 543'.0858-dc19 87-15883 CIP ©1987 Plenum Press, New York Softcover reprint of the hardcover 1s t edition 1987 A Division of Plenum Publishing Corporation 233 Spring Street, New York, N.Y. 10013 All rights reserved No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without written permission from the Publisher Contributors Hal R. Brand, Chemistry and Materials Science Department, Lawrence Livermore National Laboratory, Livermore, California J. T. Clerc, Pharmaceutical Institute, University of Bern, Bern, Switzerland Debra S. Egolf, Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania Peter R. Griffiths, Department of Chemistry, University of California, Riverside, California Stephen R. HeUer, Model and Database Coordination Laboratory, Agricultural Systems Research Institute, United States D~partment of Agriculture, Agricultural Research Service, Beltsville Agricultural Research Center, Beltsville, Maryland Gao Hongjin, Department of Chemistry and Chemical Engineering, Tsinghua University, Beijing, People's Republic of China . Peter C. Jurs, Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania David A. Laude, Jr., Department of Chemistry, University of California, Riverside, California Stephen R. Lowry, Nicolet Analytical Instrument Corporation, Madison, Wisconsin Halliday J. H. MacFie, Institute of Food Research (Bristol Laboratory), Langford, Bristol, United Kingdom Edmund R. Malinowski, Department of Chemistry and Chemical Engineering, Stevens Institute of Technology, Castle Point, Hoboken, New Jersey v vi Contributors Harald Martens, Norwegian Computing Center, Blindern, Oslo, Norway Henk L. C. Meuzelaar, Biomaterials Profiling Center, The University of Utah, Research Park, Salt Lake City, Utah Tormod Nres, Norwegian Food Research Institute, As-NLH, Norway John A. Pierce, Department of Chemistry, University of California, Riverside, California Abraham Savitzky, Silvermine Resources, Inc., Wilton, Connecticut Charles L. Wilkins, Department of Chemistry, University of California, Riverside, California Willem Windig, Biomaterials Profiling Center, The University of Utah, Research Park, Salt Lake City, Utah. Present Address: U.S. Army Chemical Research, Development and Engineering Center, Aberdeen Proving Ground, Maryland Carla M. Wong, Chemistry and Materials Science Department, Lawrence Livermore National Laboratory, Livermore, California. Present Address: NASA-Ames Research Center, Moffett Field, California Hugh B. Woodruff, Merck Sharp & Dohme Research Laboratories, Rahway, New Jersey Preface June 1986 brought together some of the world's leaders in computer enhanced analytical spectroscopy at Snowbird, Utah, for what the attendees decided to call "The First Hidden Peak Symposium." With the remarkable advances in both computer hardware and software, it is interesting to observe that, while many computational aspects of spectroscopic analysis have become routine, some of the more fundamental problems remain unsolved. The group that assembled included many of those who started trying to interpret chemical spectroscopy when computers were ponderous, slow, and not very accessible, as well as newcomers who never knew the day that spectrometers were delivered without attached computers. The synergism was excellent. Many new ideas, as well as this volume, resulted from interactions among the participants. The conclusion was that progress would be made on more fundamen tal problems now that hardware, software, and mathematics were coming together on a more sophisticated level. The feeling was that the level of sophistication is now adequate and that it is only a matter of time before automated spectral interpretation surpasses all but the most advanced human experts. We shall see. The group ended the meeting by promising to reassemble in the summer of 1988 for "The Second Hidden Peak Symposium." T. Isenhour Logan This book provides a relatively broad overview of recent advances in computerized optimization, data exploration, and spectral interpretation methods in mass spectrometry (MS), infrared spectroscopy (IR), and nuc lear magnetic resonance spectroscopy (NMR). Though many good text books on the mathematical and heuristic principles of data processing methods are available, the editors perceived the need for a text emphasizing novel applications in the rapidly expanding field of analytical spectroscopy. vii viii Preface Moreover, because of the many siIhilarities between typical -optimization, exploration, and interpretation problems encountered in different spectros copic disciplines, an attempt was made to let the scope of the book cut across the major analytical spectroscopies, namely MS, IR, and NMR. A practical consequence of this broad scope, however, was the inability to provide complete, in-depth coverage of all current topics in computer enhanced analytical spectroscopy within a single volume. Thus, selected examples were used to illustrate recent developments in several particularly active areas. In this endeavor the editors were fortunate to obtain the enthusiastic collaboration of some of the foremost experts and authorities in these areas, who agreed to contribute the twelve c.apita selecta constituting this book. Important topics that have been left uncovered or covered incom pletely include automated spectral identification and interpretation me thods in MS (for which several excellent overviews can be found in the recent literature*) and applications of multivariate analysis methods to NMR and IR data (some information can be found in Chapters 8 and 12, but most of the examples presented come from the field of MS). In trying to organize the various topics and chapters in the most consistent as well as convenient manner, it was decided to group the six chapters dealing with optimization and multivariate analysis techniques together into Part I of this book, whereas the six chapters discussing automated spectral identification and interpretation methods make up Part II. Upon studying the various contributions several novel trends in computer-enhanced analytical spectroscopy stand out, namely: (1) when used in combination, algorithmic and heuristic approaches strongly complement (rather than exclude) each other; (2) factor-analysis-based techniques are finding exciting new appli cations in enhancing the information yield from spectra of complex materials; and (3) multisource data analysis methods are showing great promise for integrating data obtained by different spectroscopic techniques. * "Mass Spectrometer Data Acquisition and Processing Systems," "Compound Identification by Computer Matching Mass Spectra," and "Use of a Computer to Identify Unknown Compounds: The Automation of Scientific Inference" in Biochemical Applications oj Mass Spectrometry, First Supplementary Volume, G. R. Waller and o. C. Dermer, eds., Wiley, New York, 1980. Preface ix In combination with the current, rapid developments in available computer hardware and software, these new trends can be expected to revolutionize the field of analytical spectroscopy over the next few years. However, rather than replacing the expert spectroscopists, computer enhanced spectroscopic techniques will greatly increase the amount of information obtainable with a given analytical instrument or set of techniques. The practical lesson for analytical spectroscopists is perhaps that it may well be more efficient to invest more time, effort, and money in the implementation of computer-enhanced optimization, deconvolution, and interpretation techniques than in the acquisition, operation, and mainten ance of ever more costly and complicated "superspectrometers." To borrow a phrase from a recent editorial in Analytical Chemistry, "math is cheaper than physics." The expert technical assistance of Melinda Van and Lexa Murphy in preparing the various chapters for publication and in completing the overall manuscript is gratefully acknowledged. Dr. Willem Windig and Joe Richards are thanked for their help and advice in reviewing and proofreading the final draft. Henk L. C. Meuzelaar Salt Lake City Contents PART I: OPTIMIZATION AND EXPLORATORY DATA ANALYSIS Chapter 1 Development of an AI-Based Optimization System for Tandem Mass Spectrometry Carla M. Wong and Hal R. Brand 1.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2. Problem Statement.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3. Proposed Method of Solution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.4. Evolution of TQMSTUNE. . . . . . . . • . . . . . . . . . . . . . . . . . • • • • . . • • . . . • • • 9 1.4.1. TQMSTUNE Version 1................................... 9 1.4.2. TQMSTUNE Version 2................................... 10 1.4.3. TQMSTUNE Version 3................................... 14 1.5. Knowledge Representation in the TQMS Domain............ 16 1.5.1. Representation of InstrumentConstruction Knowledge 17 1.5.2. Representation of Procedural Tuning Knowledge.... . 19 1.5.3. Representation of Output Evaluation Procedures..... 20 1.5.4. Representation of Interfacing Knowledge........ ... .. . 21 1.6. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1.7. Conclusion... ....... . ... .... ..................... .... ...... .. . 25 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Chapter 2 Curve Fitting and Fourier Self-Deconvolution for the Quantitative Representation of Complex Spectra Peter R. Griffiths, John A. Pierce, and Gao Hongjin 2.1. Introduction. .. .................................. .......... ... 29 2.1.1. Quantitative Analysis of Highly Overlapped Spectra. . 29 2.1.2. Derivative Spectrometry ......................... '" . . . 30 2.1.3. Fourier Self-Deconvolution. . . . . . . .. . . . .. . . . . . . . . . . . . . . 31 2.1.4. Curve-Fitting Unresolved Peaks...... .... . . ... ..... . .. 34 xi