Engineering Computational and Analytical Methodologies for Multi-Scale Biological Systems
Costly materials, technical measurement limitations, and lengthy experiments are the typical constraints on experimental approaches for quantifying biological processes. Enbisys Laboratory develops approaches based on semi-definite programming and bounded arithmetic for strategic pruning of the experimental design space to eliminate data sets that provide limited information for improving estimate quality.
System identification of biological processes is a multi-step approach that requires the integration of multiple levels of –omics data to estimate plausible structures and/or parameters capable of describing the process. Enbisys Laboratory is involved in the Systems Modeling Group for a cross-departmental collaborative research project. The objective of the project is to elucidate and model the reactions involved in lignin biosynthesis in poplar. The goal of this project is to develop hybrid mechanistic/phenomenological models that are capable of predicting changes in lignin structure with respect to transgenic perturbations.
Regulation and Modeling of Lignin Biosynthesis
Modeling of Cellulose, Hemicellulose and Lignin-Carbohydrate Complex Formation and Regulation to Understand Plant Cell Wall Structure
Dynamic Regulatory Modeling of the Iron Deficiency Response in Arabidopsis thaliana
The scale of many practical biological problems of interest is often a limiting factor in the implementation of many useful computational systems analysis approaches. This presents specific computational challenges when performing stochastic simulations or employing bounded estimation and machine learning techniques. Enbisys Laboratory develop parallel information sharing architectures that leverage the multi-core technologies found in most workstations today. This implementation utilizes POSIX threads and is based on novel combinations of mutex-locked linked lists, memory locks, and shared data, which allows each core to safely access and modify data while operating on specific tasks independently. This has allowed for significant computation savings in the previously mentioned bounded estimation algorithms.
The complete integration of models and the connection of models to phenotypic functional states is still one of the areas of systems biology that has not been fully developed. This creates an ever-widening gap between model development and true biological system design. Enbysis Laboratory is currently investigating computational approaches for compartmentalizing large-scale biological models based on the components’ relationship to functional phenotypic states. The goal is to use such approaches for multi-scale modeling and large-scale biological system design.