The application of Principal Component Analysis to materials science data

Suh, Changwon; Rajagopalan, Arun; Li, Xiang; Rajan, Krishna

doi:10.2481/dsj.1.19

Cited by 61 publications

(31 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…to differentially grown bacteria, we sought to aggregate these data into a cumulative metric of similarity among MHB-, BHI-, and M-grown bacteria. We turned to PCA, a mainstay of systems biology which is used to uncover complex patterns within large data sets (12,19,32,39). For this analysis, we only considered data for which we had sufficient matched replicates (IglB, IglC, KatG, MglB, FsaP, TNF-␣, IL-6, IL-10, IL-12, and nitrite levels in wild-type F. tularensis LVS grown in MHB and BHI and in M-grown, extracellular bacteria).…”

Section: Resultsmentioning

confidence: 99%

Adaptation of Francisella tularensis to the Mammalian Environment Is Governed by Cues Which Can Be Mimicked In Vitro

Hazlett¹,

Caldon²,

McArthur

et al. 2008

Infect Immun

Self Cite

114

View full text Add to dashboard Cite

The intracellular bacterium Francisella tularensis survives in mammals, arthropods, and freshwater amoeba. It was previously established that the conventional media used for in vitro propagation of this microbe do not yield bacteria that mimic those harvested from infected mammals; whether these in vitro-cultivated bacteria resemble arthropod-or amoeba-adapted Francisella is unknown. As a foundation for our goal of identifying F. tularensis outer membrane proteins which are expressed during mammalian infection, we first sought to identify in vitro cultivation conditions that induce the bacterium's infection-derived phenotype. We compared Francisella LVS grown in brain heart infusion broth (BHI; a standard microbiological medium rarely used in Francisella research) to that grown in Mueller-Hinton broth (MHB; the most widely used F. tularensis medium, used here as a negative control) and macrophages (a natural host cell, used here as a positive control). BHIand macrophage-grown F. tularensis cells showed similar expression of MglA-dependent and MglA-independent proteins; expression of the MglA-dependent proteins was repressed by the supraphysiological levels of free amino acids present in MHB. We observed that during macrophage infection, protein expression by intracellular bacteria differed from that by extracellular bacteria; BHI-grown bacteria mirrored the latter, while MHB-grown bacteria resembled neither. Naïve macrophages responding to BHI-and macrophage-grown bacteria produced markedly lower levels of proinflammatory mediators than those in cells exposed to MHBgrown bacteria. In contrast to MHB-grown bacteria, BHI-grown bacteria showed minimal delay during intracellular replication. Cumulatively, our findings provide compelling evidence that growth in BHI yields bacteria which recapitulate the phenotype of Francisella organisms that have emerged from macrophages.

show abstract

Section: Resultsmentioning

confidence: 99%

Adaptation of Francisella tularensis to the Mammalian Environment Is Governed by Cues Which Can Be Mimicked In Vitro

Hazlett¹,

Caldon²,

McArthur

et al. 2008

Infect Immun

Self Cite

114

View full text Add to dashboard Cite

show abstract

“…Rajan and co-workers [111][112][113][114] applied data-mining methods, such as PCA, and predictive methods, such as partial least squares (PLS) to certain fields of materials science (zeolites, semiconductors, etc.). They connected conventional materials databases with experimental data sets in searches for correlations and patterns.…”

Section: Reviewsmentioning

confidence: 99%

Combinatorial and High‐Throughput Materials Science

2007

View full text Add to dashboard Cite

There is increasing acceptance of high-throughput technologies for the discovery, development, and optimization of materials and catalysts in industry. Over the years, the relative synchronous development of technologies for parallel synthesis and characterization has been accompanied by developments in associated software and information technologies. This Review aims to provide a comprehensive overview on the state of the art of the field by selected examples. Technologies developed to aid research on complex materials are covered as well as databases, design of experiment, data-mining technologies, modeling approaches, and evolutionary strategies for development. Different methods for parallel synthesis provide single sample libraries, gradient libraries for electronic or optical materials, similar to polymers and catalysts, and products produced through formulation strategies. Many examples illustrate the variety of isolated solutions and document the barely recognized variety of new methods for the synthesis and analysis of almost any material. The Review ends with a summary of success stories and statements on still-present problems and future tasks.

show abstract

“…These compressed encodings automatically select combinations of the most informative synthesis parameters, and dimensionality reduction has been found to increase predictive performance in materials property prediction by improving the computational efficiency of training machinelearning algorithms for classification or regression. 47,48 Autoencoders are a class of neural network algorithms that learn to reproduce the identity function, and thus reconstruct the training data, while "squeezing" the data through a lowdimensional inner layer, which acts as a bottleneck. This inner layer with lower dimensionality corresponds to a continuous "latent space," which aims to preserve information from the higher-dimensional input space.…”

Section: Resultsmentioning

confidence: 99%

Virtual screening of inorganic materials synthesis parameters with deep learning

Kim

Huang

Jegelka³

et al. 2017

npj Comput Mater

171

136

View full text Add to dashboard Cite

Virtual materials screening approaches have proliferated in the past decade, driven by rapid advances in first-principles computational techniques, and machine-learning algorithms. By comparison, computationally driven materials synthesis screening is still in its infancy, and is mired by the challenges of data sparsity and data scarcity: Synthesis routes exist in a sparse, highdimensional parameter space that is difficult to optimize over directly, and, for some materials of interest, only scarce volumes of literature-reported syntheses are available. In this article, we present a framework for suggesting quantitative synthesis parameters and potential driving factors for synthesis outcomes. We use a variational autoencoder to compress sparse synthesis representations into a lower dimensional space, which is found to improve the performance of machine-learning tasks. To realize this screening framework even in cases where there are few literature data, we devise a novel data augmentation methodology that incorporates literature synthesis data from related materials systems. We apply this variational autoencoder framework to generate potential SrTiO 3 synthesis parameter sets, propose driving factors for brookite TiO 2 formation, and identify correlations between alkali-ion intercalation and MnO 2 polymorph selection.

show abstract

The application of Principal Component Analysis to materials science data

Cited by 61 publications

References 7 publications

Adaptation of Francisella tularensis to the Mammalian Environment Is Governed by Cues Which Can Be Mimicked In Vitro

Adaptation of Francisella tularensis to the Mammalian Environment Is Governed by Cues Which Can Be Mimicked In Vitro

Combinatorial and High‐Throughput Materials Science

Virtual screening of inorganic materials synthesis parameters with deep learning

Contact Info

Product

Resources

About