In recent years, the semiconductor manufacturing industry has recognized class imbalance as a major impediment to the development of high-performance fault detection (FD) models. Class imbalance ...refers to skews in class distribution in which normal wafer samples are considerably more abundant than fault samples. In such a situation, standard machine learning algorithms create FD models with classification boundaries that are biased toward majority-class data, resulting in high type II error rates. In this paper, we compare the performance of machine learning algorithms for class-imbalanced FD problems. We evaluate the performance of three sampling-based algorithms, four ensemble algorithms, four instance-based algorithms, and two support vector machine algorithms. Two experiments were conducted to compare algorithm performance using etching process data and chemical vapor deposition process data. Different data scenarios were considered by setting the imbalance ratio to three levels. The results of the experiments indicated that the instance-based algorithms presented excellent performance even when the imbalance ratio increased.
A variety of statistical and data-mining techniques have been developed for the fault detection (FD) modeling of semiconductor manufacturing processes over the past three decades. However, few ...studies have analyzed which models are adequate for different types of fault data. In this paper, we define a FD model as an algorithm combining feature extraction, feature selection, and classification. We prepare six process data scenarios and collect data by simulating an etching tool. In total, 117 possible algorithm combinations are tested as FD models for the six datasets. With these test results, we conduct statistical analyses from two perspectives: 1) the algorithm perspective and 2) FD model perspective. From the algorithm perspective, we compare the performance of competing algorithms in the three model-building steps using multiple comparison methods and discuss the advantages and disadvantages of individual algorithms. From the model perspective, we determine which algorithm combinations are recommended for FD models of the semiconductor process and explain why some combinations do not exhibit the expected performance. In both analyses, we interpret some results using 3-D plots.
Training fault detection model requires advanced data-mining algorithms when the growth rate of the process data is notably high and normal-class data overwhelm fault-class data in number. Most ...standard classification algorithms, such as support vector machines (SVMs), can handle moderate sizes of training data and assume balanced class distributions. When the class sizes are highly imbalanced, the standard algorithms tend to strongly favor the majority class and provide a notably low detection of the minority class as a result. In this paper, we propose an online fault detection algorithm based on incremental clustering. The algorithm accurately finds wafer faults even in severe class distribution skews and efficiently processes massive sensor data in terms of reductions in the required storage. We tested our algorithm on illustrative examples and an industrial example. The algorithm performed well with the illustrative examples that included imbalanced class distributions of Gaussian and non-Gaussian types and process drifts. In the industrial example, which simulated real data from a plasma etcher, we verified that the performance of the algorithm was better than that of the standard SVM, one-class SVM and three instance-based fault detection algorithms that are typically used in the literature.
Single-molecule fluorescence is widely used to study conformational complexity in proteins, and has proven especially valuable with intrinsically disordered proteins (IDPs). Protein studies using ...dual-color single-molecule Förster resonance energy transfer (smFRET) are now quite common, but many could benefit from simultaneous measurement of multiple distances through multi-color labeling. Such studies, however, have suffered from limitations in site-specific incorporation of more than two dyes per polypeptide. Here we present a fully site-specific three-color labeling scheme for α-synuclein, an IDP with important putative functions and links to Parkinson disease. The convergent synthesis combines native chemical ligation with regiospecific cysteine protection of expressed protein fragments to permit highly controlled labeling via standard cysteine-maleimide chemistry, enabling more global smFRET studies. Furthermore, this modular approach is generally compatible with recombinant proteins and expandable to accommodate even more complex experiments, such as by labeling with additional colors.
Display omitted
•Site-specific labeling of a single polypeptide is presented for three-color smFRET•Convergent method combines regiospecific protection and native chemical ligation•Full-length protein is synthesized from expressed and labeled fragments•Modular and expandable method enables more colors and combined biophysical study
Lee and Moran et al. present a strategy for fully site-specific labeling of the intrinsically disordered protein α-synuclein with three fluorophores for multi-color single-molecule FRET. This expandable method uses standard maleimide-cysteine chemistry, and enables more global investigation of complex proteins and their dynamic interactions.
NK cells respond to tumor and virus-infected cells directly through several activation receptors, including natural cytotoxicity receptors, or indirectly through the activating Fc receptor CD16 for ...antibody-coated cells. Triggering of NK-cell effector functions through these receptors depends on physically associated transmembrane signaling adaptors, such as FcRγ (also known as FcεRIγ) and CD3ζ, both of which have been traditionally believed to be expressed by all mature NK cells. However, we have identified a distinct subset of human NK cells that are deficient for FcRγ expression but express normal levels of CD3ζ. FcRγ-deficient NK cells were readily detectable in about one-third of the healthy individuals examined. The deficiency was confined to the CD56(dim) population and was due to low FcRγ mRNA. FcRγ-deficient NK cells displayed dramatically reduced expression of the natural cytotoxicity receptors NKp46 and NKp30 but still expressed substantial levels of CD16. Compared to FcRγ-expressing NK cells, FcRγ-deficient NK cells showed poor direct reactivity toward tumor targets as measured by cytokine production and degranulation. Unexpectedly, however, FcRγ-deficient NK cells exhibited significantly more robust responsiveness upon stimulation through CD16, particularly for cytokine production, compared to FcRγ-expressing NK cells. Thus, our study reveals FcRγ-deficient NK cells as a novel subset of human NK cells that have remarkably potent responses toward antibody-coated targets. These findings also illustrate a differential contribution of FcRγ and CD3ζ for the expression and functional activity of their associated receptors.
In the last decade, a wide range of avian influenza viruses (AIVs) have infected various mammalian hosts and continuously threaten both human and animal health. It is a result of overcoming the ...inter-species barrier which is mostly associated with gene reassortment and accumulation of mutations in their gene segments. Several recent studies have shed insights into the phenotypic and genetic changes that are involved in the interspecies transmission of AIVs. These studies have a major focus on transmission from avian to mammalian species due to the high zoonotic potential of the viruses. As more mammalian species have been infected with these viruses, there is higher risk of genetic evolution of these viruses that may lead to the next human pandemic which represents and raises public health concern. Thus, understanding the mechanism of interspecies transmission and molecular determinants through which the emerging AIVs can acquire the ability to transmit to humans and other mammals is an important key in evaluating the potential risk caused by AIVs among humans. Here, we summarize previous and recent studies on molecular markers that are specifically involved in the transmission of avian-derived influenza viruses to various mammalian hosts including humans, pigs, horses, dogs, and marine mammals.
Consolidating multiple servers into a physical machine is now a commonplace in cloud infrastructures. The virtualized systems often arrange virtual disks of multiple virtual machines (VMs) on the ...same underlying storage device while striving to guarantee the service level objective (SLO) of the performance of each VM. Unfortunately, sync operations called by a VM may make it hard to satisfy the performance SLO by disturbing I/O activities of other VMs. In this paper, we experimentally uncover that the disk cache flush operation incurs significant I/O interference among VMs, and revisit the internal architecture and flush mechanism of the flash memory-based SSD. Then, we present vFLUSH , a novel VM-aware flush mechanism, that supports VM-based persistency control for the disk cache flush operation. We also discuss the long-tail latency issue in vFLUSH and an efficient scheme for mitigating the problem. Our evaluation with various micro- and macro-benchmarks shows that vFLUSH reduces the average latency of disk cache flush operations by up to 58.5%, thereby producing improvements in throughput by up to <inline-formula> <tex-math notation="LaTeX">1.93\times </tex-math></inline-formula>. The method for alleviating the long-tail latency problem, which is applied to vFLUSH , achieves a significant reduction in tail latency by up to 75.9%, with a modest throughput degradation by 2.9-7.2%.
Total internal reflection fluorescence-based single-molecule Förster resonance energy transfer (FRET) measurements were previously carried out on the ankyrin repeat domain (ARD) of IκBα, the ...temporally regulated inhibitor of canonical NFκB signaling. Under native conditions, most of the IκBα molecules showed stable, high FRET signals consistent with distances between the fluorophores estimated from the crystal structures of the NFκB(RelA/p50)-IκBα complex. Similar high FRET efficiencies were found when the IκBα molecules were either free or in complex with NFκB(RelA/p50), and were interpreted as being consistent with the crystallographically observed ARD structure. An exception to this was observed when the donor and acceptor fluorophores were attached in AR3 (residue 166) and AR6 (residue 262). Surprisingly, the FRET efficiency was lower for the bound IκBα molecules (0.67) than for the free IκBα molecules (0.74), apparently indicating that binding of NFκB(RelA/p50) stretches the ARD of IκBα. Here, we conducted confocal-based single-molecule FRET studies to investigate this phenomenon in greater detail. The results not only recapitulated the apparent stretching of the ARD but also showed that the effect was more pronounced when the N-terminal domains (NTDs) of both RelA and p50 were present, even though the interface between NFκB(RelA/p50) and IκBα encompasses only the dimerization domains. We also performed mass spectrometry-detected amide hydrogen/deuterium exchange (HDXMS) experiments on IκBα as well as IκBα bound to dimerization-domain-only constructs or full-length NFκB(RelA/p50). Although we expected the stretched IκBα to have regions with increased exchange, instead the HDXMS experiments showed decreases in exchange in AR3 and AR6 that were more pronounced when the NFκB NTDs were present. Simulations of the interaction recapitulated the increased distance between residues 166 and 262, and also provide a plausible mechanism for a twisting of the IκBα ARD induced by interactions of the IκBα proline-glutamate-serine-threonine-rich sequence with positively charged residues in the RelA NTD.
Many cellular functions are critically dependent on the folding of complex multimeric proteins, such as p97, a hexameric multidomain AAA+ chaperone. Given the complex architecture of p97, ...single‐molecule (sm) FRET would be a powerful tool for studying folding while avoiding ensemble averaging. However, dual site‐specific labeling of such a large protein for smFRET is a significant challenge. Here, we address this issue by using bioorthogonal azide–alkyne chemistry to attach an smFRET dye pair to site‐specifically incorporated unnatural amino acids, allowing us to generate p97 variants reporting on inter‐ or intradomain structural features. An initial proof‐of‐principle set of smFRET results demonstrated the strengths of this labeling method. Our results highlight this as a powerful tool for structural studies of p97 and other large protein machines.
Watching one p97 at a time: The single‐molecule biophysical chemistry of p97, a complex, cysteine‐rich, essential cellular protein machine, can now be studied using smFRET, thanks to the use of advanced site‐specific dye labeling by an unnatural amino acid and click chemistry.