Overview
The following sections show some solutions to some common problems revolving around the lack of uniformity of gene name nomenclature, and show examples of Ontologizer output for C. elegans, mouse, yeast and Drosophila.Gene Ontology Files
The Ontologizer requires the Gene Ontology Terms file (gene_ontology.obo), which is the file that describes individual GO terms and their relationships to one another. This file can be downloaded from the Gene Ontology homepage. Alternatively, the Ontologizer can automatically download the latest version if you specify the URL to this file.Gene Association Files
The OBO file describes only the functional hierarchy of terms and does not provide functional annotations for actual genes. For this purpose, users need to download the appropriate annotation file. These files are generally entitled gene_association.XXX, where XXX stands for the database or species (e.g., mgi for Mouse Genome Informatics, and sgd for Saccharomyces Genome Database. Association files can also be downloaded from the Gene Ontology homepage. The Ontologizer can also automatically download the latest version of the association file for a number of commonly analyzed organisms (See the tutorial within the Ontologizer application for more details).Data Files
The Ontologizer produces listings of GO annotations for user supplied lists of genes or gene products. One situation in which this can be useful is for "clustering" analysis of microarray data, but there are many other potential uses. The Ontologizer assumes that each group of genes resides in its own file, and presently accepts FASTA files as well as files in which each gene is on its own line. For FASTA and plain files, the first word on a line (for FASTA files, the first word following the ">" sign) is taken to be the name of the gene or gene product, and anything else on the line is taken to be a description. It is easy to extend the source code to allow for other types of file formats.
The user should copy files representing the results of clustering into a separate directory. Note that the names of the genes or gene products need to correspond to the nomenclature used in the association file for the Ontologizer to function properly. This can be especially problematic if (for instance) EST names are present in the results of clustering. Several solutions to such problems are presented below on this web page in the form of Perl scripts that can transform files and gene names such that the Ontologizer will recognize the names.
Downloading the Ontologizer Source Code
The source code contains all the code as well as an Eclipse project file. The source code can be imported as an Eclipse project and compiled with Eclipse.
Preparing Data for the Ontologizer
The Ontologizer uses the gene or gene-product nomeclature as found in the various gene_association files. The Ontologizer recognizes names entered as DB_Object, DB_Object_Name, or Synonym in these files. However, in many cases, gene names will be listed in other form, such as for instance EST accession numbers or Affymetrix ids. Further details on the gene_association files can be found on the Gene Ontology website. In these cases, it will be necessary to transform your data by mapping non-standard gene ids to a form that will be recognized by the Ontologizer.
Although there are many ways of doing this, we recommend using scripts to transform the files into new files where the first word is the standard gene or gene-product name, and everything after the first tab-stop is an optional description consisting of at least the original name or accession number (so that the user can refer back to the original data if desired). The Ontologizer will automatically output the description underneath the standard name in the "detail" page for each cluster.
The following sections present sample output from the Ontologizer for several different species. In many cases, we were able to use the output files directly without further processing (For C. elegans for instance). In other cases, it was necessary to create a mapping for the names of the original files such that they would conform to those used in the gene_association files. We present some "cookbook" Perl scripts that demonstrate how to perform such transformations.
C. elegans
The recommended nomeclature for C. elegans forsees that genes are given names consisting of three italicized letters, a hyphen, and an italicized Arabic number, e.g., dpy-5. However, it is still not uncommon to see cosmid names: Sequences that are predicted to be genes from sequence data alone were initially named on the basis of the sequenced cosmid, plus a number. For example, the genes predicted for the cosmid K07F5 are called K07F5.1, K07F5.2, and so on. As more becomes known, it is expected that these names will be replaced by standard three-letter names.In this example, we used the Ontologizer to analyze a set of genes identified by Liu et al. (2004) as being regulated by TGFbeta during entry into dauer diapause in C. elegans.
When resources are scant, C. elegans larvae arrest as long-lived dauers under the control of insulin/IGF- and TGFbeta-related signaling pathways. The authors identified genes that show different levels of expression in a comparison of wild-type L2 or L3 larvae (non-dauer) to TGFbeta mutants at similar developmental stages undergoing dauer formation. The data shown here represent the strongly regulated (>2.1 fold) genes in this experiment and were taken from table 2. The population set was taken to be all C. elegans genes for which annotations were provided by Wormbase. Among other things, GO analysis revealed that a number of terms related to ribosomes were highly significantly overrepresented. The gene names in these clusters represent a mix of cosmid and standard names. The gene_association.wb file provided by Wormbase and the cluster files with gene names were used as is. The sample input file and population file can be downloaded here:
Mus musculus
Mouse Genome Informatics (MG) (Jackson Laboratories) supplies a gene_association file for M. musculus. They also provide guidelines for mouse gene nomeclature. The gene_association.mgi file provided by MGI contain annotations based primarily on MGI accession numbers. Therefore, one way of using the Ontologizer to analyze mouse datasets that are based on NCBI accession numbers is to transform the accession numbers into the corresponding MGI accession numbers. We present a way of doing this with perl scripts using files from NCBI's Entrez Gene (gene_info and gene2accession).The perl script NCBI2MG.pl maps NCBI accession numbers via Entrez Gene ids to MGI accession numbers. The output file, NCBI2MGI.txt, can be used as a mapping file for the Ontologizer. The original accession numbers are visible (together with the full gene name, locusID, and UniGene cluster) directly after the MGI accession number. For accession numbers for which no MGI id could be identified (mainly some ESTs), just the original accession number is displayed. The corresponding original files can be downloaded here: