UNLABELLED: Correspondence analysis of codon usage data is a widely used method in sequence analysis, but the variability in amino acid composition between proteins is a confounding factor when one wants to analyse synonymous codon usage variability. A simple and natural way to cope with this problem is to use within-group correspondence analysis. There is, however, no user-friendly implementation of this method available for genomic studies. Our motivation was to provide to the community a Web facility to easily study synonymous codon usage on a subset of data available in public genomic databases. AVAILABILITY: Availability through the Pole Bioinformatique Lyonnais (PBIL) Web server at http://pbil.univ-lyon1.fr/datasets/charif04/ with a demo allowing us to reproduce the figure in the present application note. All underlying software is distributed under a GPL licence. CONTACT: http://pbil.univ-lyon1.fr/members/lobry.

Summary: MADE4, microarray ade4, is a software package that facilitates multivariate analysis of microarray gene-expression data. MADE4 accepts a wide variety of gene-expression data formats. MADE4 takes advantage of the extensive multivariate statistical and graphical functions in the R package ade4, extending these for application to microarray data. In addition, MADE4 provides new graphical and visualization tools that aid in interpretation of multivariate analysis of microarray data. Availability: The R package MADE4 is available from Bioconductor http://bioinf.vcd.ie/software and from Bioconductor http://www.bioconductor.org Contact: [email protected] Supplementary information: MADE4 is well documented. There are tutorials, in the form of vignettes, which describe typical analyses. In addition, the MADE4 manual provides descriptions and examples for each function.

This report describes two applications of a multivariate method for studying classes of nucleotide or protein sequences: correspondence discriminant analysis (CDA). The first example is the discrimination between Escherichia coli proteins according to their subcellular location (membrane, cytoplasm and periplasm). The high resolution of the method made it possible to predict the subcellular location of E.coli proteins for whom this information is not known. The second example is discrimination between the coding sequences of leading and lagging strands in four bacteria: Mycoplasma genitalium, Haemophilus influenzae, E.coli and Bacillus subtilis. The programs used for computing the analysis are integrated in a publicly available package that runs on MacOS 7.x or Windows 95 operating systems (http:/(/)biomserv.univ-lyonl.fr/ADE-4.html). These programs are also accessible through our World Wide Web server (http:/(/)biomserv.univ-lyonl.fr/Net Mul.html).

We have developed a World-Wide Web server for browsing sequence collections structured under the ACNUC format and for performing multivariate analyses on sequences. General collections (like GenBank or EMBL), as well as specialized data banks (like Hovergen and NRSub) can be accessed. This system allows complex queries to be constructed, and the result of each query, represented by a list of sequences, is stored on the server. It is then possible to reuse this list to compute multivariate analyses on the sequences. Two examples of applications are shown. The first one consists in a study of codon usage with correspondence analysis on all the protein genes of Haemophilus influenzae Rd. This study allows the highly expressed genes and the integral membrane proteins of this organism to be identified. The second one consists in an ordering of 70 aligned protein sequences of growth hormone with principal coordinate analysis. With this method, we are able to re-establish the patterns of relationships between the sequences previously determined with tree building programs.

A multivariate analysis method called co-inertia analysis was used to determine the main relationships between two data tables having identical rows. This method is available in the ADE multivariate analysis package for Macintosh micro-computers. It was applied to two data sets, one containing the amino-acid composition of 999 E. coli proteins, and the other the values of 402 physico-chemical properties for the 20 natural amino-acids. There were strong relationships between amino-acid physico-chemical properties and the composition of proteins. The first common factor was hydrophobicity; it is linked to the biological environment of proteins, either in the cytoplasm (or outside the cell), or in the nonpolar environment of the phospholipid bilayer of biological membranes. The second factor linked the expressivity of protein genes and the propensity of amino-acids to form alpha helix/beta sheets. The third factor showed that heavy, aromatic amino-acids tend to be avoided, except when they are needed for structural or functional reasons. These results are discussed in terms of selective pressure acting on amino-acid composition of proteins.

Two Macintosh programs written for multivariate data analysis and multivariate data graphical display are presented. MacMul includes principal component analysis (PCA), correspondence analysis (CA) and multiple correspondence analysis (MCA), with a complete, original and unified set of numerical aids to interpretation. GraphMu is designed for drawing collections of elementary graphics (curves, maps, graphical models) thus allowing comparisons between variables, individuals, and principal axes planes of multivariate methods. Both programs are self-documented applications and make full use of the user-oriented graphical interface of the Macintosh to simplify the process of analysing data sets. An example is described to show the results obtained on a small ecological data set.