Figure 1.
Example of Shannon's entropy calculation for a gene with four splicing isoforms SP1‥SP4.
EST counts are provided for each isoform in a normal and cancer tissue. In this example, isoform entropy is higher in the cancer tissue (1.38 versus 1.16 bits).
Figure 2.
Ratio of average isoform entropy in cancer versus normal tissues.
A value of 1 indicates that average entropy per gene in cancer tissue = average entropy per gene in normal tissue. The first number in parentheses corresponds to the number of genes that were used to calculate entropy gains, and the second corresponds the total coverage in ESTs/cDNAs/SAGEs for the diseased and normal tissue types. Only tissue types for which at least 50 genes and 100 isoforms were available to measure the entropy ratio are shown. (A) alternative initiation. (B) Alternative polyadenylation. (C) Alternative splicing.
Table 1.
Gene Ontology term biases for genes with entropy gain in high-entropy cancer tissues, as measured using the Gene Ontology Toolbox [38].
Figure 3.
Meta-analysis method to obtain proliferative indices of cancer samples in microarray experiments.
The 188 genes of the “cell cycle” cluster in the conserved coexpression network identified by Stuart et al. [26] were extracted. Each of the 3787 cancer-related samples was classified in one of 5 separate bins of same size in function of the average expression level of these 188 genes. The high proliferation signature bin (High PS) corresponds to the 20% of samples that have the highest mean expression level of the 188 genes; the lowest proliferation signature bin (Low PS) corresponds to the 20% of samples that have the lowest mean expression level of the 188 genes.
Figure 4.
Correlation between the proliferation signature of different cancers and their splicing entropy ratio.
Figure 5.
Ratio of average isoform entropy in fetus versus adult tissues for alternative splicing.
The first number in parentheses corresponds to the number of genes that were used to calculate entropy gains, and the second corresponds to the total coverage in ESTs/cDNAs for the fetal and adult tissue types.
Figure 6.
Classification of cancer-specific splice events as proposed by Xu and Lee [29].
Three typical cases of cancer-specific events are shown. Numbers are EST counts supporting each splice form. S: putative normal splice form; S': putative cancer-specific splice form. Percentages in parenthesis indicate the proportion of overall cancer-specific events that belong to each category according to [29].
Figure 7.
Models for mechanisms leading to specific or non-specific expression of splice isoforms in cancer tissues.
Dotted arrows: hypothetical links. Box A: Known trans effect in which change in splice factor activation results in specific changes in the expression levels of several splice variants. Box B: Possible alternative mechanism in which disruption of SR protein splicing induces a wider deregulation of splice isoform expression. The dotted arrow between boxes indicates a possible link between specific and non-specific splicing disruption that may occur preferentially in proliferating tumors.