GOurmet: a tool for quantitative comparison and visualization of gene expression profiles based on gene ontology (GO) distributions

Abstract

Background: The ever-expanding population of gene expression profiles (EPs) from specified cells and tissues under a variety of experimental conditions is an important but difficult resource for investigators to utilize effectively. Software tools have been recently developed to use the distribution of gene ontology (GO) terms associated with the genes in an EP to identify specific biological functions or processes that are over- or under-represented in that EP relative to other EPs. Additionally, it is possible to use the distribution of GO terms inherent to each EP to relate that EP as a whole to other EPs. Because GO term annotation is organized in a tree-like cascade of variable granularity, this approach allows the user to relate (e.g., by hierarchical clustering) EPs of varying length and from different platforms (e.g., GeneChip, SAGE, EST library).

Results: Here we present GOurmet, a software package that calculates the distribution of GO terms represented by the genes in an individual expression profile (EP), clusters multiple EPs based on these integrated GO term distributions, and provides users several tools to visualize and compare EPs. GOurmet is particularly useful in meta-analysis to examine EPs of specified cell types (e.g., tissue-specific stem cells) that are obtained through different experimental procedures. GOurmet also introduces a new tool, the Targetoid plot, which allows users to dynamically render the multi-dimensional relationships among individual elements in any clustering analysis. The Targetoid plotting tool allows users to select any element as the center of the plot, and the program will then represent all other elements in the cluster as a function of similarity to the selected central element.

Conclusion: GOurmet is a user-friendly, GUI-based software package that greatly facilitates analysis of results generated by multiple EPs. The clustering analysis features a dynamic targetoid plot that is generalizable for use with any clustering application.

Figure 1

The GOurmet Vocabulary program calculates and displays the fractional representation of every GO term in each inputted expression profile. Depicted is a screen capture of the GUI showing the fractional representation of GO terms in a sample gene expression profile (in this case, the list of genes preferentially expressed in gastric zymogenic cells). Other profiles can be selected by clicking on the tabs. Results can be outputted in a multi-sheet MS Excel Workbook file (not depicted) for each expression profile, where one sheet summarizes the statistics of the annotation (number of genes in the profile, percent of those genes successfully annotated with GO terms, etc.), the second sheet lists every GO term associated with each gene, the third lists all the GO terms found in the profile, followed by the genes associated with each GO term. Results can also be output as tab-delimited Comparison files as input for GO Cartography.

Figure 2

The GOurmet Cartography dendrogram and wave plot windows. Bottom left – the GOCART dendrogram plots all the expression profiles in a hierarchically clustered dendrogram using a modified Pearson's correlation to determine distances between clusters. A sliding bar allows the user to select how sub-clusters are colorized to visualize differences. The dendrogram can be exported as a jpeg. In the example, note how the three sample progenitor expression profiles (MeSC [32], NSC, and GEP) cluster together, and three of the four sample differentiated cell lineage expression profiles cluster together (mature blood, brain, and zymogenic cells [33]). Parietal cells, a highly specialized, differentiated, mitochondria-rich cell type cluster separately, well away from the other lineages. Numbers in brackets indicate reference from which each expression profile was acquired. Top – the GOCART wave diagram allows users to select GO terms (listed at the left of the window) to view their fractional representation among selected expression profiles (profiles to be viewed are selected from the dendrogram plot). Each GO term is automatically assigned a different color in the plot. In the example, note how much higher the fractional representation of the GO term "mitochondrion" (red line) is in parietal cells relative to the other sample expression profiles. The wave diagram can be output as a jpeg image.

Figure 3

GOurmet can cluster expression profiles from varying platforms, from different labs, and of different lengths. A) Three different hematopoietic stem cell (HSC) expression profiles (one generated from a subtracted library [16], two from different GeneChips in different labs [5,6]) all cluster most closely with each other (marked by arrowheads) than with any of the other progenitor or differentiated cell profiles. B) A parent gene list composed of every annotated gene on the Moe430V2.0 GeneChip was reduced in length 10-fold by random exclusion of 9 in 10 genes and 100-fold by exclusion of 99 in 100 genes. The reduced length versions of the original list still cluster with the parent (marked by arrowheads). [32]

Figure 4

The GOurmet Cartography Targetoid plotting window. Depicted is a screen shot where the sample GEP profile has been chosen as a center element (center elements are selectable in the dendrogram window), and all other profiles radiate outward in direct inverse proportion to how similar they are to GEPs. Note how GEPs are more similar to the other two stem/progenitor profiles and more distant from all the profiles of differentiated cells. Relationships among non-center elements are only approximately reflective of their similarity and are depicted by how far apart they are from each other (θ angle between elements in this polar coordinate system). Any profile can be selected as the center and, by multiply selecting different profiles, users can reconstruct all the direct relationships among the various profiles. A sliding bar allows the user to focus on certain regions of the targetoid space (important for visualization when large numbers of profiles are compared at once and many similar ones cluster closely together); users can also adjust the axial scale. The targetoid plot can be output as a jpeg.

Figure 5

The GOurmet Cartography Cartesian plotting window. This window toggles with the Targetoid plot. Here, users can select one GO term as the y-axis and one as the x-axis. All the expression profiles are plotted according to their fractional representation of the selected GO terms. Note how the two selected GO terms ("nucleus" and "integral to membrane") distinguish expression profiles of stem/progenitor cells as a group from those of differentiated cells.