nci60 {robustHD} | R Documentation |
NCI-60 cancer cell panel
Description
The data set is a pre-processed version of the NCI-60 cancer cell panel as used in Alfons, Croux & Gelper (2013). One observation was removed since all values in the gene expression data were missing.
Usage
data("nci60")
Format
Protein and gene expression data on 59 observations are stored in two separate matrices:
protein
a matrix containing protein expressions based on antibodies (162 columns), acquired via reverse-phase protein lysate arrays and log2 transformed.
gene
a matrix containing gene expression data (22283 columns), obtained with an Affymetrix HG-U133A chip and normalized with the GCRMA method.
In addition, meta information on the proteins, genes, and cancer cell lines is stored in three separate data frames:
proteinInfo
a data frame with 162 rows and the following 4 columns:
Experiment
(the name of the experiment for collecting the data),Probe
(the name of the individual probe),Symbol
(the symbol of the protein in Human Genome Organisation (HUGO) nomenclature), andID
(identifier of the protein per the National Center for Biotechnology Information (NCBI) Entrez database). The rows of this data frame correspond to the columns of the matrixprotein
.geneInfo
a data frame with 22283 rows and the following 4 columns:
Experiment
(the name of the experiment for collecting the data),Probe
(the name of the individual probe),Symbol
(the symbol of the gene in Human Genome Organisation (HUGO) nomenclature), andID
(identifier of the gene per the National Center for Biotechnology Information (NCBI) Entrez database). The rows of this data frame correspond to the columns of the matrixgene
.cellLineInfo
a data frame with 59 rows and 15 columns containing various information on the cancer cell lines, such as tissue of origin and histology, or age and sex of the patient. The rows of this data frame correspond to the rows of the matrices
protein
andgene
.
Source
The original data were downloaded from https://discover.nci.nih.gov/cellminer/ on 2012-01-27.
The exact version of the data used in Alfons, Croux & Gelper (2013) can be
obtained from https://github.com/aalfons/nci60, together with the
script for pre-processing. The data in package robustHD differ in
that the matrix of the gene expressions is called gene
and that they
include the three data frames with meta information on proteins, genes, and
cancer cell lines.
References
Reinhold, W.C., Sunshine, M., Liu, H., Varma, S., Kohn, K.W., Morris, J., Doroshow, J. and Pommier, Y. (2012) CellMiner: A Web-Based Suite of Genomic and Pharmacologic Tools to Explore Transcript and Drug Patterns in the NCI-60 Cell Line Set. Cancer Research, 72(14), 3499–3511. doi:10.1158/0008-5472.CAN-12-1370
Alfons, A., Croux, C. and Gelper, S. (2013) Sparse least trimmed squares regression for analyzing high-dimensional large data sets. The Annals of Applied Statistics, 7(1), 226–248. doi:10.1214/12-AOAS575
Examples
# load data
data("nci60")
# define response variable
y <- protein[, 92]
# screen most correlated predictor variables
correlations <- apply(gene, 2, corHuber, y)
keep <- partialOrder(abs(correlations), 100, decreasing = TRUE)
X <- gene[, keep]