Examining large databases: A chemometric approach using principal component analysis

Meglen, Robert R.

doi:10.1002/cem.1180050305

Cited by 64 publications

(45 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…PCA score plotting is a very good tool for identifying outliers, influential points, and data entry errors. 60 Figure 3A shows that 7 of the 25 026 compounds appeared at the extremes. By checking the original database, we found that the GI 50 values of the four compounds (NSC numbers 619989, 620130, 624589, and 626674) located at the upper left corner of Figure 3A were simply miscoded in the database: the exponents of concentrations that should have been -4 or -6 were miscoded as +4 or +6.…”

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

“…Meglen has given an excellent description of the use of PCA to examine large databases. 60 Koutsoukos et al 21 have used PCA to analyze the NCI cell screen data for a set of 141 "standard" anticancer agents for which the mechanisms of action are well-defined. 20 PCA score plots showed distinct clusters of compounds for some of the mechanisms of action examined.…”

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

“…21,60 Compounds similar in activity pattern should group together on the score plots, whereas dissimilar compounds should scatter over the plots. A particular example is the PCA analysis on a data set of 279 analogues of ellipticine, ellipticinium, and camptothecin.…”

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

See 2 more Smart Citations

Mining and Visualizing Large Anticancer Drug Discovery Databases

Shi

Fan

Lee

et al. 1999

J. Chem. Inf. Comput. Sci.

View full text Add to dashboard Cite

In order to find more effective anticancer drugs, the U.S. National Cancer Institute (NCI) screens a large number of compounds in vitro against 60 human cancer cell lines from different organs of origin. About 70 000 compounds have been tested in the program since 1990, and each tested compound can be characterized by a vector (i.e., "fingerprint") of 60 anticancer activity, or -[log(GI 50 )], values. GI 50 is the concentration required to inhibit cell growth by 50% compared with untreated controls. Although cell growth inhibitory activity for a single cell line is not very informative, activity patterns across the 60 cell lines can provide incisive information on the mechanisms of action of screened compounds and also on molecular targets and modulators of activity within the cancer cells. Various statistical and artificial intelligence methods, including principal component analysis, hierarchical cluster analysis, stepwise linear regression, multidimensional scaling, neural network modeling, and genetic function approximation, among others, can be used to analyze this large activity database. Mining the database can provide useful information: (a) for the development of anticancer drugs; (b) for a better understanding of the molecular pharmacology of cancer; and (c) for improvement of the drug discovery process.

show abstract

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

Section: Principal Component Analysis (Pca)mentioning

confidence: 99%

See 1 more Smart Citation

Mining and Visualizing Large Anticancer Drug Discovery Databases

Shi

Fan

Lee

et al. 1999

J. Chem. Inf. Comput. Sci.

View full text Add to dashboard Cite

show abstract

“…Here it is demonstrated how the reduction of uncorrelated noise provided by principal component analysis (PCA) can be exploited for automated alignment. PCA is an unsupervised multivariate technique that uses orthogonal linear transformations of the data onto a new coordinate system to project the greatest variance of the data onto the first component, the second greatest variance onto the second component, and so forth [11][12][13]. This processing extracts features with less uncorrelated noise from the datasets, thus allowing their automated alignment.…”

mentioning

confidence: 99%

Automated, feature-based image alignment for high-resolution imaging mass spectrometry of large biological samples

Broersen

Liere

Altelaar

et al. 2008

J. Am. Soc. Mass Spectrom.

View full text Add to dashboard Cite

High-resolution imaging mass spectrometry of large biological samples is the goal of several research groups. In mosaic imaging, the most common method, the large sample is divided into a mosaic of small areas that are then analyzed with high resolution. Here we present an automated alignment routine that uses principal component analysis to reduce the uncorrelated noise in the imaging datasets, which previously obstructed automated image alignment. An additional signal quality metric ensures that only those regions with sufficient signal quality are considered. We demonstrate that this algorithm provides superior alignment performance than manual stitching and can be used to automatically align large imaging mass spectrometry datasets comprising many individual mosaic tiles. (J Am Soc Mass Spectrom 2008, 19, 823-832) © 2008 American Society for Mass Spectrometry I maging mass spectrometry is a rapidly developing analytical tool because it provides the ability to map the profiles of specific biomolecules, in which the intrinsic mass of the molecule differentiates between any modified forms; to record the distributions of multiple analytes in parallel; and to perform these analyses without a label and with clinical samples [1]. This combination of specificity, parallel detection, and non-targeted analysis has led to great excitement for its potential as a discovery tool.It is the goal of many research groups to be able to perform high-resolution analysis of large samples, thus combining the ability to examine distributions between organs/tumors and their surroundings as well as to investigate the subcellular/intracellular locations of the biomolecules. The central premise of this approach is that the subcellular locations will provide some of the information required to explain differences in the more global patterns.High spatial resolution measurements are a wellestablished ability of secondary ion mass spectrometry (SIMS). Recent advances in both ionization efficiency (polyatomic primary ions) and sample preparation have significantly improved the sensitivity for detecting intact, medium-sized molecular ions (Ͻ1000 Da) from tissues and cells [1][2][3][4][5][6]. High-resolution images of small peptides, lipids, cholesterol, vitamins, and pharmaceuticals have all been reported, and through the use of large polyatomic primary ions three-dimensional (3D) molecular imaging results are beginning to appear [7][8][9].The images are normally created by moving the ionization beam in a set pattern across the sample and performing mass analysis at each point of the raster (spatially correlated mass spectrometry). The raster pattern typically uses 8-or 10-bit encoding; as a result the maximum analysis field contains 256 ϫ 256 (or 1024 ϫ 1024) pixels. In a typical high spatial resolution SIMS measurement the pixel size is about 200 nm, meaning maximum analysis areas of approximately 50 or 200 m, respectively. This is much smaller than many of the biological samples of interest; for example, a tissue section of an adult rat ...

show abstract

“…The PCA and CA employed correlation ( =0.05) matrices on the variables in order to establish possible associations and input sources among polluting elements as described by Delgado, Nieto et al (2010). In PCA, the eigenvalues of the principal components are a measure of their associated variances (Meglen, 1992;Mellinger, 1987;Wenning and Erickson, 1994). Correlation of principal components and original variables is given by loadings.…”

Section: Discussionmentioning

confidence: 99%

Assessing water quality in lake Naivasha

Ndungu¹

View full text Add to dashboard Cite

Examining large databases: A chemometric approach using principal component analysis

Cited by 64 publications

References 6 publications

Mining and Visualizing Large Anticancer Drug Discovery Databases

Mining and Visualizing Large Anticancer Drug Discovery Databases

Automated, feature-based image alignment for high-resolution imaging mass spectrometry of large biological samples

Assessing water quality in lake Naivasha

Contact Info

Product

Resources

About