Genstruct - Genstruct Technology

Genstruct Technology

Genstruct has built a technology platform that leverages the world of known biology combined with large panomic datasets to identify molecular mechanisms of disease and drug action.

This platform takes advantage of advances in knowledge theory and systems engineering to build a broad knowledgebase through the Knowledge Assembly™ process. This platform continuously incorporates specie and tissue specific biological knowledge into a central large, unified computable knowledgebase that can be used to generate Causal System Models and formulate testable hypotheses of disease and drug action.

The technology to mine the literature and data, to incorporate knowledge and to test and validate models against experimental data enables Genstruct to develop highly complex and accurate knowledgebase for specific biological systems. The computable format of this knowledgebase allows the creation, testing and validation of hypotheses through algorithmic implementation of artificial intelligence logic. The end result is the rigorous development and evaluation of hypotheses about mechanisms, markers and effects that can be directly tested against additional large-scale experimental data or through subsequent validation experiments.

Genstruct's platform combines causal models that capture known information about causal relationships between biomolecules with our proprietary platform that automates reasoning. Causal Knowledge Assembly Models are attractive because they support diverse computational approaches to large-scale, automated reasoning and inferencing. At the core of the Genstruct platform is a logical modeling framework that infers causes for changes in the system’s components as it transitions between states (e.g., what are the potential causes for RNA expression state changes between two timepoints) and models effects on the components resulting from system perturbations.

This type of causal modeling is qualitative in nature, in that it describes the direction of influence on the abundance or activities of genes, proteins and metabolites. This approach is a best of breed hybrid between more abstract association models that cannot be used to infer causes or effects and more detailed quantitative models that can be used to infer causes and effects but require very detailed quantitative knowledge (such as diffusion or rate constants that are very difficult or even impossible to obtain).

A key feature of the Genstruct platform is the ability for logical simulation. It is possible to take the observed changes in the dataset and ask what events could lead to the observed changes. This is called a reverse causal simulation, and it works upstream from the observed changes through the biological networks to identify control points. Conversely, the platform is also equipt to perform forward causal simulation by perturbing a particular control point and identifying all the changes that would be expected downstream from this perturbation. Examples of both simulation approaches are illustrated in the figure below.

This automated reasoning system within the Genstruct platform facilitates the development of biologically-driven insights and enables a complete exploration of the possible explanations of the gene, protein and metabolite changes observed in an experiment. The methods are based on logical scientific reasoning and produce results that are readily testable and verifiable by laboratory scientists. This platform is designed to focused on finding causal relationships that unite the observed changes in an experiment, as opposed to statistical and correlative techniques that rely on coincident observations to infer conclusions.