Edward V. Thomas

Multivariate classification of infrared spectra of cell and tissue samples

Infrared microspectroscopy of biopsied canine lymph cells and tissue was performed to investigate the possibility of using IR spectra coupled with multivariate classification methods to classify the samples as normal, hyperplastic, or neoplastic (malignant). IR spectra were obtained in transmission mode through BaF2 windows and in reflection mode from samples prepared on gold-coated microscope slides. Cytology and histopathology samples were prepared by a variety of methods to identify the optimal methods of sample preparation. Cytospinning procedures that yielded a monolayer of cells on the BaF2 windows produced a limited set of IR transmission spectra. These transmission spectra were converted to absorbance and formed the basis for a classification rule that yielded 100% correct classification in a cross-validated context. Classifications of normal, hyperplastic, and neoplastic cell sample spectra were achieved by using both partial least-squares (PLS) and principal component regression (PCR) classification methods. Linear discriminant analysis applied to principal components obtained from the spectral data yielded a small number of misclassifications. PLS weight loading vectors yield valuable qualitative insight into the molecular changes that are responsible for the success of the infrared classification. These successful classification results show promise for assisting pathologists in the diagnosis of cell types and offer future potential for in vivo IR detection of some types of cancer.

Release of extraction-resistant mRNA in stationary phase Saccharomyces cerevisiae produces a massive increase in transcript abundance in response to stress

BackgroundAs carbon sources are exhausted, Saccharomyces cerevisiae cells exhibit reduced metabolic activity and cultures enter the stationary phase. We asked whether cells in stationary phase cultures respond to additional stress at the level of transcript abundance.ResultsMicroarrays were used to quantify changes in transcript abundance in cells from stationary phase cultures in response to stress. More than 800 mRNAs increased in abundance by one minute after oxidative stress. A significant number of these mRNAs encode proteins involved in stress responses. We tested whether mRNA increases were due to new transcription, rapid poly-adenylation of message (which would not be detected by microarrays), or potential release of mature mRNA present in the cell but resistant to extraction during RNA isolation. Examination of the response to oxidative stress in an RNA polymerase II mutant, rpb1-1, suggested that new transcription was not required. Quantitative RT-PCR analysis of a subset of these transcripts further suggested that the transcripts present in isolated total RNA from stationary phase cultures were polyadenylated. In contrast, over 2,000 transcripts increased after protease treatment of cell-free lysates from stationary phase but not exponentially growing cultures. Different subsets of transcripts were released by oxidative stress and temperature upshift, suggesting that mRNA release is stress-specific.ConclusionsCells in stationary phase cultures contain a large number of extraction-resistant mRNAs in a protease-labile, rapidly releasable form. The transcript release appears to be stress-specific. We hypothesize that these transcripts are associated with P-bodies.

Development of Robust Multivariate Calibration Models

Multivariate calibration is frequently used for the quantitative analysis of a wide variety of materials using spectroscopy in the agricultural and food industries, manufacturing industries, medical sciences, and pharmaceutical industries. To date, most of the research activities in the multivariate calibration literature have focused on data analysis for model building and prediction with methods such as principal-components regression or partial least squares regression. As an alternative to focusing on data-analytic activities, we consider the ability of an experimental design to improve the robustness of the resulting calibration model. Through an example involving diffuse reflectance measurements, we illustrate how consideration of environmental and instrumental factors during the experimental design phase can result in a calibration model that is robust (against natural environmental and instrument variations) and easy to maintain. In this example, the analyte of interest produces a spectroscopic signal that is very weak in comparison to the environmental and instrumental factors.

Errors-in-variables estimation in multivariate calibration

A set of q responses, y = (y 1, y 2, …, y q,) T , is related to a set of p explanatory variables, x = (x 1, x 2, …, x p ) T , through the classical linear regression model, y T = a T + x T B + e T . First, the unknown parameters a and B are estimated using a calibrafion set. The statistical problem that is considered here is that of estimating the vector x o, that underlies a new observed vector of responses y o using the parameter estimates obtained from the first procedure. These two procedures are commonly referred to as calibration and prediction (or inverse prediction) and sometimes jointly referred to as calibration. The prediction procedure can be viewed as parameter estimation in errors-in-variables regression. The maximum likelihood estimator (assuming normally distributed measurement errors) is proposed for the prediction procedure. Unlike the classical estimator used in the prediction procedure, the proposed estimator is consistent with respect to the number of response variables. The performan...

Thermal Study of Inverter Components

Thermal histories of inverter components were collected from operating inverters from several manufacturers and three locations. The data were analyzed to determine thermal profiles, and the dependence on local conditions, as well as to assess the effect on inverter reliability. Inverter temperatures were shown to increase with the power dissipation of the inverters, follow diurnal and annual cycles, and have a dependence on wind speed. An accumulated damage model was applied to the temperature profiles, and an example of using these data to predict reliability was explored.

Statistical analysis of microarray data with replicated spots : a case study with Synechococcus WH8102

Until recently microarray experiments often involved relatively few arrays with only a single representation of each gene on each array. A complete genome microarray with multiple spots per gene (spread out spatially across the array) was developed in order to compare the gene expression of a marine cyanobacterium and a knockout mutant strain in a defined artificial seawater medium. Statistical methods were developed for analysis in the special situation of this case study where there is gene replication within an array and where relatively few arrays are used, which can be the case with current array technology. Due in part to the replication within an array, it was possible to detect very small changes in the levels of expression between the wild type and mutant strains. One interesting biological outcome of this experiment is the indication of the extent to which the phosphorus regulatory system of this cyanobacterium affects the expression of multiple genes beyond those strictly involved in phosphorus acquisition.

Evaluating the Ignition Sensitivity of Thermal-Battery Heat Pellets

Thermal batteries are activated by the ignition of heat pellets. If the heat pellets are not sensitive enough to the ignition stimulus, the thermal battery will not activate, resulting in a dud. Thus, to assure reliable thermal batteries, it is important to demonstrate that the pellets have satisfactory ignition sensitivity by testing multiple specimens. There are many statistical methods for evaluating the sensitivity of a device to some stimulus. Generally, these methods are applicable to the situation in which a single test is destructive to the specimen being tested, independent of the outcome of the test. In the case of thermal-battery heat pellets, however, tests that result in a nonresponse do not totally degrade the specimen. This peculiarity provides opportunities to efficiently evaluate the ignition sensitivity of heat pellets. In this article, a simple strategy for evaluating heat-pellet ignition sensitivity (including experimental design and data analysis) is described. The relatively good asy...

Improvements in Methods for Spectral Combination of Gas Chromatography/Fourier Transform Infrared Spectroscopic Data

Coaddition of spectra in a single-component peak of a gas Chromatograph (GC) obtained with a Fourier transform infrared spectrometer is the method generally used to improve the signal-to-noise ratio (S/N) of the spectrum of the eluted analyte. It is commonly thought that coaddition of spectra to a relative intensity level of 40% of the GC peak will lead to the optimal improvement in S/N of the resulting composite spectrum. We have shown that this is not generally the case for either simulated Gaussian-shaped or experimentally obtained asymmetric GC bands. The optimal intensity level for coaddition is found to be a function of the shape of the GC band and the ratio of the number of background to sample scans used in generating the individual IR spectra. We have also introduced the use of classical least-squares (CLS) techniques as a superior method to improve the S/N of the composite analyte spectrum. With the use of CLS methods, spectra included in generating the composite spectrum can be a small fraction of the maximum intensity in the GC peak while still resulting in S/N improvements. The theoretical S/N of the composite spectrum with the use of CLS methods is shown to be always as good as or better than that achieved with the coaddition method. The improvements achieved in S/N when CLS methods are used can be more than a factor of two greater than results for the traditional coaddition method for the cases considered in this paper. Furthermore, it is shown that increasing the number of background to sample scans is a very convenient method to improve the S/N of the composite spectrum obtained by either method. The results presented here for GC/FT-IR are also generally applicable to LC/FT-IR, SFC/FT-IR, and TGA/FT-IR for bands that contain a single analyte.

Studies of high reliability long-life Li/SO2 cells

Abstract To satisfy the need for a 5-year battery, a long-term test of Li/SO 2 cells was initiated. Premature failures were observed and a program was undertaken to identify the cause of these failures. Based on the results of this program, changes in the cell design were made which gave rise to the Sandia modified Li/SO 2 cell having improved long-life performance. Several batches of modified cells have been manufactured during the past five years. Acceptance testing of incoming cells from each batch revealed batch-to-batch variation in performance as well as a small proportion of cells with low capacity. A program to develop a non-destructive methodology for identifying cells with low inherent capacity was initiated. Several relevant non-destructive measurement parameters were identified. A decision tree screening technique, based on these parameters, was developed to identify those cells having low capacity.

Statistical Analysis for Nuclear Forensics Experiments

As with any type of forensics, nuclear forensics seeks to infer historical information using models and data. This article connects nuclear forensics and calibration. We present statistical analyses of a calibration experiment that connect several responses to the associated set of input values and then ‘make a measurement’ using the calibration model. Previous and upcoming real experiments involving production of PuO2 powder motivate this article. Both frequentist and Bayesian approaches are considered, and we report findings from a simulation study that compares different analysis methods for different underlying responses between inputs and responses, different numbers of responses, different amounts of natural variability, and replicated or non-replicated calibration experiments and new measurements. Published 2015. This article is a U.S. Government work and is in the public domain in the USA. Statistical Analysis and Data Mining: The ASA Data Science Journal, 2015