Q. Peter He

Self-Calibrated Warping for Mass Spectra Alignment

With recent advances in mass spectrometry (MS) technologies, it is now possible to study protein profiles over a wide range of molecular weights in small biological specimens. However, MS spectra are usually not aligned or synchronized between samples. To ensure the consistency of the subsequent analysis, spectrum alignment is necessary to align the spectra such that the same biological entity would show up at the same m/z value for different samples. Although a variety of alignment algorithms have been proposed in the past, most of them are developed based on chromatographic data and do not address some of the unique characteristics of the serum or other body fluid MS data. In this work, we propose a self-calibrated warping (SCW) algorithm to address some of the challenges associated with serum MS data alignment. In addition, we compare the proposed algorithm with five existing representative alignment methods using a clinical surface enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF-MS) data set.

Comprehensive evaluation of two genome‐scale metabolic network models for Scheffersomyces stipitis

Genome‐scale metabolic network models represent the link between the genotype and phenotype of the organism, which are usually reconstructed based on the genome sequence annotation and relevant biochemical and physiological information. These models provide a holistic view of the organisms metabolism, and constraint‐based metabolic flux analysis methods have been used extensively to study genome‐scale cellular metabolic networks. It is clear that the quality of the metabolic network model determines the outcome of the application. Therefore, it is critically important to determine the accuracy of a genome‐scale model in describing the cellular metabolism of the modeled strain. However, because of the model complexity, which results in a system with very high degree of freedom, a good agreement between measured and computed substrate uptake rates and product secretion rates is not sufficient to guarantee the predictive capability of the model. To address this challenge, in this work we present a novel system identification based framework to extract the qualitative biological knowledge embedded in the quantitative simulation results from the metabolic network models. The extracted knowledge can serve two purposes: model validation during model development phase, which is the focus of this work, and knowledge discovery once the model is validated. This framework bridges the gap between the large amount of numerical results generated from genome‐scale models and the knowledge that can be easily understood by biologists. The effectiveness of the proposed framework is demonstrated by its application to the analysis of two recently published genome‐scale models of Scheffersomyces stipitis. Biotechnol. Bioeng. 2015;112: 1250–1262.

Control performance assessment and diagnosis for semiconductor processes

In this work we briefly review the current status of control performance assessment (CPA) and control performance diagnosis (CPD) in semiconductor processes, and present an integrated CPA/CPD framework for semiconductor processes. By identifying the model-plant mismatch and process disturbance dynamics explicitly, CPA and CPD can be performed simultaneously in the proposed framework. Two performance indices and an on-line implementation scheme are introduced, and the performance of the proposed CPA/CPD methods is demonstrated using two simulation studies.

Elucidating xylose metabolism of scheffersomyces stipitis by integrating principal component analysis with flux balance analysis

The conversion of pentose to ethanol is one of the major barriers of industrializing lignocellulosic ethanol processes. As the most promising native strain for pentose fermentation, Scheffersomyces stipitis (formerly known as Pichia stipitis) has been widely studied for its xylose fermentation. In spite of the abundant experimental evidence regarding ethanol and by-products production under various aeration conditions, the mathematical descriptions of the processes are rare. In this work, the constraint-based metabolic network model for the central carbon metabolism of S. stipitis was reconstructed by integrating genomic (P. stipitis v2.0, KEGG), biochemical (ChEBI, KEGG) and physiological information available for this microorganisms and other related yeast. The model consists of the stoichiometry of metabolic reactions, the biosynthetic requirements for growth and other constraints. Flux Balance Analysis is applied to characterize the phenotypic behavior of S. stipitis grown on xylose. The model predictions are in good agreement with published experimental results. In addition, a series of specially designed in silico experiments are performed, and PCA has been applied to analyze the results to elucidate the redox balance of S. stipitis for xylose fermentation. The analysis revealed key metabolic reactions related to redox homeostasis and could provide important insights on cofactor engineering of xylose metabolism.

A Comprehensive Evaluation of Statistics Pattern Analysis Based Process Monitoring

Abstract Statistics pattern analysis (SPA) is a new multivariate statistical monitoring framework proposed by the authors recently. It addresses some challenges that cannot be readily addressed by the commonly used multivariate statistical methods such as principal component analysis (PCA) in monitoring batch processes in the semiconductor industry. It was later extended to the monitoring of continuous processes using a moving window based approach. In this work, we perform a comprehensive comparison of SPA with representative linear and nonlinear multivariate process monitoring methods. The superior performance of SPA is demonstrated using the challenging Tennessee Eastman process (TEP).

An Alternative Stiction-modelling Approach and Comparison of Different Stiction Models

Due to the adverse impact of valve stiction in the process industry, both physical models and empirical data-driven models have been developed in the past decade to investigate the valve-stiction behaviour. Chapter 2 presented two data-driven models: Choudhury’s model and Kano’s model. In this chapter, we focus on another data-driven modelling approach (i.e. He’s model).We compare different data-driven models against a well-established physical model. The comparison is carried out in several aspects, including model assumptions, valve signatures generated by the models, and closed-loop behaviour when valve models are included in a closed-loop system. A new data-driven model is proposed based on a thorough analysis of the physical model and its effectiveness in simulating valve stiction is demonstrated.

Statistics pattern analysis based virtual metrology for plasma etch processes

Virtual metrology (VM) is the prediction of end-of-batch properties (i.e., metrology data) using process variables and other information available for the process and/or the product (i.e., machine data) without physically conducting property measurement. VM (sometimes augmented with existing metrology) has been utilized in semiconductor process monitoring and control. Besides the economic benefit of replacing or reducing metrology tools, due to the instant availability of high frequency machine data, a good VM can actually provide better process monitoring and control performance compared to the same monitoring and control schemes based on the physical metrology data which often obtained at lower frequencies and usually with delays. In this paper, we propose a statistics pattern analysis (SPA) based VM approach for predicting sheet resistance using optical emission spectroscopy (OES) data. The advantageous properties of the SPA based VM are discussed. And the performance of the SPA based VM is compared with several commonly used VM algorithms in terms of prediction accuracy.

Comparison of a new spectrum alignment algorithm with other methods

Many data generated by the analytical techniques (such as chromatography, mass spectrometry (MS), near infrared (NIR) spectroscopy and nuclear magnetic resonance (NMR) spectroscopy) are not synchronized or aligned, where the same peak may show up at different positions for different samples. Before applying any analysis method to these spectra, data synchronization/alignment is needed to ensure that the same variable represents the same attribute in all samples. In this work, we propose a three-step self-calibrated warping (SCW) algorithm and apply it to align a clinical prostate cancer data set generated by mass spectrometry. The performance of SCW, in terms of alignment quality and computation speed, is compared with two existing methods. In addition, the robustness of SCW to its tuning parameters and data noise is studied.

A Novel Soft Sensor Approach for Estimating Individual Biomass in Mixed Cultures

Due to many advantages associated with mixed cultures, their application in biotechnology has expanded rapidly in recent years. At the same time, many challenges remain for effective mixed culture applications. One obstacle is how to efficiently and accurately monitor the individual cell populations. Current approaches on individual cell mass quantification are suitable for off‐line, infrequent characterization. In this study, we propose a fast and accurate “soft sensor” approach for estimating individual cell concentrations in mixed cultures. The proposed approach utilizes optical density scanning spectrum of a mixed culture sample measured by a spectrophotometer over a range of wavelengths. A multivariate linear regression method, partial least squares or PLS, is applied to correlate individual cell concentrations to the spectrum. Three experimental case studies are used to examine the performance of the proposed soft sensor approach.

Quantification of valve stiction based on a semi-physical model

Valve stiction is one of the most common equipment problems that can cause poor performance in control loops. Consequently, there is a strong need in the process industry for non-invasive methods that can not only detect but also quantify stiction. In this work, a semi-physical valve stiction model is derived from the analysis of the dynamic response of a physical model. Based on the semi-physical model, we propose a noninvasive valve stiction quantification method using the routine operating data from the process. The algorithm is proposed to estimate the stiction parameters, namely static friction and dynamic or kinetic friction, without requiring the valve position signal. Quantification is accomplished by using linear and nonlinear least-squares methods which are robust and easy to implement. Several simulation examples, including both self-regulating and integrating processes with different degrees of stiction, are used to demonstrate the effectiveness of the method.