Microarrays are widely used to assess mRNA expression profiles on a genome-wide scale, producing a large amount of data.
The analysis of these data can be confusing for non specialist users due to the increasing number of methodologies and tools currently available.
Based on the experience of biostatisticians of Institut Curie, we propose both a clear analysis protocol and the tools to investigate the data. It provides a useful starting point for many microarrays users.
The most usual and relevant existing R functions were discussed, validated and gathered in an easy-to-use R package (EMA) devoted to Affymetrix GeneChip analysis. These functions were improved in order to facilitate use, visualization and interpretation of results.
The current version of EMA offers an entire analysis strategy including :
Normalisation of Affymetrix arrays
Exploratory analysis (Clustering, Principal Component Analysis, Exploratory plots)
Differential Analysis (Student-based tests, SAM, ANOVA, linear model, Gene Set Analysis)
The EMA package is linked to many other CRAN and BioConductor
packages. Some of the existing functions were extended or
re-implemented in a easy-to-use way.
the follwing packages are required for the EMA installation :
cluster
heatmap.plus
FactoMineR
GOstats
survival
multtest
affy
gcrma
GSA
RankProd
siggenes
MASS
hgu133plus2.db
xtable
biomaRt
The package can be installed using the following command line :
The help pages of all the EMA function are available here.
Contacts
The members of the EMA working group are pleased to answer any question or address any concerns you may have with the EMA package. Please send your
request at :
EMA is a free package under GPL licence. However, if
you use the package in your work please cite the following
paper :
Servant, N, Gravier, E, Gestraud, P, Laurent, C,
Paccard, C, Biton, A, Brito, I, Mandel, J, Asselain,
B, Barillot, E, Hupe, P (2010) EMA
- A R package for Easy Microarray data
analysis. BMC Res Notes, 3:277.
Acknowledgements
We thank all the persons involved in this project (P. Neuvial, S. Carpentier, J. Trolet, F. Valet) as well as the members of the ABCIS methodology group for their collaboration and fruitful discussions.
This work was supported by a grant from the Institut Curie "PIC Bioinformatique et Biostatistique" incentive and collaborative program.
