The solution for making sense of large complex data sets is to use an integrative genomic tool. Researchers at the NIH developed GRANITE (Genetic Regulatory Analysis of Networks Investigational Tool Environment) as a software workbench for representing, combining and interpreting biological models, particularly network models. GRANITE allows related models, for example, patient versus control or responder versus non-responder, to be composed using the logical operators AND/NOT/OR. Each unique combination of these operators results in a distinct way to partition the set of all available relationships into those that are part of the network and those that are not. For instance, responder AND non-responder yields the ‘common’ or ‘intersection’ network made up of the relationships that are common to both responder and non-responder groups. GRANITE supports six different methods to partition a graph using these logical operators:
- Responder: relationships significant to the responder group;
- Non-responder: relationships significant to the non-responder group;
- Responder-only: responder AND NOT non-responder;
- Non-responder-only: NOT responder AND non-responder;
- Common network: responder AND non-responder;
- Union network: responder OR non-responder;
The exclusion networks, responder-only and non-responder-only, are often the most useful because they capture relationships that are unique to treatment responders or non-responders, making those relationships particularly interesting research targets.
Availability – This tool is open source and is available for use and collaboration (the link to the code is: http://exon.niaid.nih.gov/granite/).