qCABIN: Tool for Systems Analysis of Biological Networks

The wealth and complexity of data generated due to recent technological advances demands the development of new computational methodologies to organize the experimental data/observations into quantitative models.  Indeed, the new field of Systems Biology has arisen, having as its goal the understanding of the basic networks, dynamic processes, feedback control loops, and signaling mechanisms underlying cellular processes. Current approaches for high-throughput data (in particular gene expression) analysis, such as time series analysis, pattern discovery, clustering, and class prediction, are primarily data-centric and designed to identify correlative or causal relationships signifying disease with low robustness. In contrast our qCABIN analysis protocols consider the data in the biological context of a “pathway” model.

QCABIN provides a clear framework for:

Our development effort focuses on three distinct areas

Integrative Data Management & Bioinformatic Analysis

The primary objective of our effort in this area is to design and develop a database management system (DBMS) supporting high-throughput and high-content biological investigations of today. The DBMS will manage diverse data types (microarray, spectral, ELISA) and support data storage/retrieval/querying with links to relevant online databases. Direct benefits from the proposed DBMS and systems biology-centered assay design methodologies include:

Network Collocation & Boolean Analysis

Collocation: We have automated the complex process of placing induced/repressed genes in the context of known canonical pathways and later concatenate, prune and curate the individual pathways together into a comprehensive pathway.

Boolean Analysis: We have also developed an adaptation of Boolean Network Dynamics for the analysis of these assembled, comprehensive networks.  Based on the topology of the network and without requiring often unavailable kinetics, the algorithm objectively ranks the most critical interactions within the network model with substantially minimized computational costs compared to other current methods. This feature enhances our ability to examine critically needed larger, complex networks and avoid traps of oversimplification.  

Model Identification, Parameter Estimation & Dynamic Analysis

Mechanistic methods, in contrast with informatic approaches, seek to uncover “cause-effect” relationships (rather than correlative) and establish quantitative behavior rules among the stimulus and response descriptors.  Mechanistic modeling of complex biochemical systems can be classified into three distinct methodologies: Stoichiometric, Kinetic/Deterministic, and Kinetic/Stochastic.  While kinetic models hold the ultimate promise of a deterministic description of cellular behavior, there is a unique and complex challenge associated with identifying robust models from limited data and estimation of parameters to complete the specification of the model.