Determination of strongly overlapping signaling activity from microarray data

Abstract

Background: As numerous diseases involve errors in signal transduction, modern therapeutics often target proteins involved in cellular signaling. Interpretation of the activity of signaling pathways during disease development or therapeutic intervention would assist in drug development, design of therapy, and target identification. Microarrays provide a global measure of cellular response, however linking these responses to signaling pathways requires an analytic approach tuned to the underlying biology. An ongoing issue in pattern recognition in microarrays has been how to determine the number of patterns (or clusters) to use for data interpretation, and this is a critical issue as measures of statistical significance in gene ontology or pathways rely on proper separation of genes into groups.

Results: Here we introduce a method relying on gene annotation coupled to decompositional analysis of global gene expression data that allows us to estimate specific activity on strongly coupled signaling pathways and, in some cases, activity of specific signaling proteins. We demonstrate the technique using the Rosetta yeast deletion mutant data set, decompositional analysis by Bayesian Decomposition, and annotation analysis using ClutrFree. We determined from measurements of gene persistence in patterns across multiple potential dimensionalities that 15 basis vectors provides the correct dimensionality for interpreting the data. Using gene ontology and data on gene regulation in the Saccharomyces Genome Database, we identified the transcriptional signatures of several cellular processes in yeast, including cell wall creation, ribosomal disruption, chemical blocking of protein synthesis, and, critically, individual signatures of the strongly coupled mating and filamentation pathways.

Conclusion: This works demonstrates that microarray data can provide downstream indicators of pathway activity either through use of gene ontology or transcription factor databases. This can be used to investigate the specificity and success of targeted therapeutics as well as to elucidate signaling activity in normal and disease processes.

Figure 1

Data analysis flowchart. The data was downloaded from Rosetta Inpharmatics and filtered to include only genes and experiments that showed significant variation. Bayesian Decomposition analysis generated patterns and associated gene lists for all dimensionalities between 3 and 25. ClutrFree was used to interpret these results, including use of the MIPS database of ontologies.

Figure 2

The average persistence across all dimensions. The average persistence across the dimensions is plotted for 3 to 25 dimensions. The significant drop between 15 and 16 dimensions suggests that 15 patterns provides the correct dimensionality for analysis.

Figure 3

Yeast MAPK signaling for mating and filamentation. The strongly linked MAPK signaling pathways for mating and filamentation are shown schematically with black arrows indicating mating pathway signaling and gray arrows showing filamentation pathway signaling. The mating pathway is initiated by binding to Ste2p or Ste3p receptors, while the causative molecular trigger for filamentation is unclear. The pathways share many components.

Figure 4

Relationship of patterns across dimensionalities. The results for all patterns identified in all runs of Bayesian Decomposition are summarized here. The top row shows three patterns from an analysis with 3 dimensions, while the bottom row shows 25 dimensions. The highlighted node is pattern 13 in 15 dimensions, which is the pattern identified as the mating response. Nodes are connected as described in the text using Pearson correlation measures. The numbers within the nodes are indices and have no intrinsic meaning. Each number provides the row index for P and column index for A for the analysis at that level.

Figure 5

A sample calculation of the average persistence for a single node. The average persistence is calculated by comparing the persistence at each node in the tree given in Figure 4. Each assignment of each mutant (4 are shown here) to a pattern is binarized as described in the text, then the average persistence for a node is calculated by checking on the number of times the mutant assigned to the pattern occurs in the connected nodes. The mutant can occur in any branch below the node of interest to be considered as present. If it occurs in multiple child nodes at a single level, that is still treated as a single occurrence for that level. The average for a dimension is then the average of the persistence of all nodes at that level.