Using mechanistic Bayesian networks to identify downstream targets of the sonic hedgehog pathway

Abstract

Background: The topology of a biological pathway provides clues as to how a pathway operates, but rationally using this topology information with observed gene expression data remains a challenge.

Results: We introduce a new general-purpose analytic method called Mechanistic Bayesian Networks (MBNs) that allows for the integration of gene expression data and known constraints within a signal or regulatory pathway to predict new downstream pathway targets. The MBN framework is implemented in an open-source Bayesian network learning package, the Python Environment for Bayesian Learning (PEBL). We demonstrate how MBNs can be used by modeling the early steps of the sonic hedgehog pathway using gene expression data from different developmental stages and genetic backgrounds in mouse. Using the MBN approach we are able to automatically identify many of the known downstream targets of the hedgehog pathway such as Gas1 and Gli1, along with a short list of likely targets such as Mig12.

Conclusions: The MBN approach shown here can easily be extended to other pathways and data types to yield a more mechanistic framework for learning genetic regulatory models.

Figure 1

BN and MBN modeling of Shh pathway: (a) A sequential BN model of the pathway that omits protein activity, (b) a parallel BN model of the pathway, and (c) an MBN model of the pathway. The ovals Shh, Ptch1 and Smo represent mRNA measurements, while the ovals, ShhP, Ptch1P and SmoP, represent proteins. The oval GeneX indicates a candidate downstream expression target of the pathway. Dotted ovals represent entities that are known in the mechanism but are not observed, while unbroken ovals represent experimentally observed variables. Arrows between nodes represent a directional interaction but do not specify the functional form.

Figure 2

Comparing the relative gene expression profile of 8 selected putative targets. From these expression patterns one can see that the profiles for Gas1 and Gli1 closely follow changes in Shh and Ptch1 expression. Gli2 and Gli3 show higher activation in the adult tissues as compared to the somite samples but, even in the adult tissues, their expression pattern does not correspond to the knockout state of Shh. Foxa2 expression shows a strong response to the Ptch1 knockouts in the somite samples and with Shh knockouts in the adult head tissues but there is no pattern in the adult limb and trunk tissues. While the patterns in Gli2, Gli3, Foxa2 are significant, the patterns either do not coincide with patterns in the early steps of the Shh pathway (Shh, Ptch1, Smo) indicating that there may be other genes involved in their regulatory control or the patterns do not hold over all tissues. Accordingly, those genes do not score highly in the MBN method.