Supplemental Data for:
A gene expression map for C. elegans, Science, 293: 2087-2092. 2001. (full text, abstract)
Stuart K. Kim, Jim Lund, Moni Kiraly, Kyle Duke, Min Jiang, Joshua M. Stuart, Andreas Eizinger, Brian N. Wylie, and George S. Davidson
Figure 1. A. Pie chart shows types of experiments used to generate the gene expression terrain map. Numbers in parentheses refer to the number of microarray hybridizations done for that experiment class, out of a total of 553 different microarray hybridizations. Some microarray hybridizations fall into multiple classes. B. Construction of the gene expression terrain map by VxInsight". Expression data involving 17,661 genes and 553 experiments are shown. Yellow denotes increased relative gene expression and blue denotes decreased gene expression. Only 3 genes and 3 experiments are shown for simplicity. The expression data are used to calculate Pearson correlations between every pairwise combination of genes. The most correlated genes in the correlation matrix are used to construct a two-dimensional scatterplot. The scatterplot is converted to a gene expression terrain map showing the gene correlations in three dimensions, with the altitude of a mountain corresponding to density of the genes.
Figure 2 A. C. elegans gene expression map created by VxInsight" at lowest resolution. Three dimensional representation of 44 gene mountains representing 17,661 genes and 553 microarray hybridizations. Some of the gene classes that are enriched in specific mountains are shown. B. Terrain map derived from randomized data. Additional validation experiments are discussed below. Also, there a movie flying around the topomap (85Mb or 6 Mb) and a movie zooming into the central mountain (33Mb or 3Mb). C, D. Correlation of VxInsight mountains with biological functions. This figure shows that the gene mountains formed by VxInsight (based on expression data) correlate with sets of genes that have similar biochemical functions. The correlation is much more than would be expected by random chance, indicating that the enrichments of certain types of genes in specific expression mountains are not random noise.
Figure 3. A. Mount 4 (sperm). Sperm-enriched and MSP genes are shown in red and green, respectively. B. Enlarged view of MSP genes (green) and sperm-enriched genes (red) in mount 4. C. Germ line genes in mounts 7, 11, 18 and 20. Sperm-enriched (green), oocyte-enriched (blue) and germ line-enriched genes (red) from Reinke et al. are shown. Numbers refer to mountains. D. Mount 8 (Intestine). Intestinal (green) and protease (blue) genes are shown. E. Mount 16 (muscle). Muscle (blue) and collagen (green) genes are shown. F. Mount 26 (male). Male-enriched (green) and lectins (blue) are shown. Movies of mount 4 (54Mb or 4Mb), germ line mountains (65Mb or 5Mb), mount 16 (34Mb or 2Mb).
Figure 4 A. Heat shock induction profiles for 10 genes in mount 36. Heat shocks were for 15 minutes at 33 °C, and RNA expression levels were measured 30 minutes after heat shock. Results show average expression levels (+/- SE) from four independent experiments. Black indicates genes that encode heat shock proteins and red indicates genes not previously known to be regulated by heat shock. 5B. Transposon expression in males and sperm (expanded). Genes in each of the major classes of transposon are shown. male/herm. refers to experiments comparing adult male to adult hermaphrodite RNAs. High sperm/oocyte refers to experiments comparing fem-3(gf) to fem-1(lf) worms. Yellow and blue denote high and low expression levels, respectively. C. Transposons. Tc1 (red), Tc3 (blue) and Mariner (yellow) transposons are indicated. Numbers refer to mountains. Movie of collagen mountains (71Mb or 5Mb) and transposon mountains.
Validation of VxInsight Clustering Algorithm
Link 1. Information about VxInsight, and how the gene expression topomap is derived.
Link 2. Path dependence: Shows the stability of the gene expression terrain map to starting from random points in the data.
Link 3. Noise dependence: Shows stability after a small amount of random noise is added to the data.
Link 4. Experiment dependence: Shows how many genes have similar neighbors when the set of experiments are split into two, and then used to create two independent gene expression terrain maps.
Searching and viewing the C. elegans gene expression terrain map
Supplemental Table 1: lists of genes in each mountain and for different classes discussed in the paper. Also lets you compare sets of genes to each other. There are 4 ways to view the expression data from the paper.
Supplemental Table 2: list of biogroups used in Fig. 2C and 2D.
Supplemental Table 3: list of gene expression mountains in Fig. 2C and 2D.
The best way to view the data from this paper is to download VxInsight to view the gene expression database. VxInsight will run on Windows NT, 2000 and ME, but will not run on Windows 98 or Macintosh. With VxInsight, you can view the topomap from any position and from any magnification. You can search and plot out specific sets of genes. There is also an instructional video on how to use VxInsight. (The video does not play on all PCs, and may require a decompression application. We are working on the video so that it will work on more PCs.)
Another way to view the data is to use these plotting programs. The first program allows you to search and plot out specific sets of genes on a 2D scatterplot. The second program let you find genes that are co-expressed with your favorite gene.
A third way to view the data is to download a table showing the X Y coordinates of the genes. Then plot the data from this table in a spreadsheet such as Microsoft Excel.
The fourth program allows you to see the published expression data for a set of genes.
Gene expression data
This is a file containing the raw expression data for the entire expression database. It includes all of the expression data from all of the experiments. The experiments are listed with arbitrary names so that unpublished data remains anonymous. The data are shown as log2(expression ratio).