(72a) The Dionesus Algorithm Provides Scalable and Accurate Reconstruction of Biological Networks to Reveal New Drug Targets
- Conference: AIChE Annual Meeting
- Year: 2015
- Proceeding: 2015 AIChE Annual Meeting Proceedings
- Group: Emerging Frontiers in Systems and Synthetic Biology
- Time: Monday, November 9, 2015 - 8:30am-8:48am
The DIONESUS algorithm provides scalable and accurate reconstruction of biological networks to reveal new drug targets.
Mark F. Ciaccio and Neda Bagheri. Department of Chemical & Biological Engineering, McCormick School of Engineering, Northwestern University.
Introduction: Advances in genomic and proteomic technology brings with it an unprecedented breadth of data on the molecular markers driving cellular function. With this new age in biomedical engineering comes the pressing need to find efficient strategies to glean deeper mechanistic understanding of biological processes. Scalable and rapid algorithms are critical in pushing forward innovation at the interface of medicine, mathematics, and biology. To this end, we created a scalable algorithm, termed DIONESUS, based on partial least squares regression with variable selection to reconstruct networks from large matrices of cue-signal-response relationships. Our scalable algorithm performs with higher specificity and sensitivity than several top-performing network inference algorithms in a fraction of the computational time. We applied DIONESUS to reconstruct the signaling network of a model epidermoid cancer cell line and identified STAT1 as a central modulator of apoptosis specifically in cells that have hyperactivity of the EGF receptor. Background: Deriving correlations between biomolecules, such as RNA expression or protein abundance, and cell phenotype can suggest factors that are associated with disease. Deriving the underlying network structure can identify disease drivers by elucidating control structures such as feedback loops and redundant pathways. Several algorithms can reconstruct directed biological networks using information theory, decision trees, or Bayesian networking; algorithms with high accuracy, however, are often limited to small networks . Ideally, network reconstruction can be expanded to genomic or proteomic proportions in order to broaden and deepen understanding of the dysregulation of biological systems in pathological states. Materials and Methods: After being validated in three in silico networks, DIONESUS was applied to a high confidence dataset characterizing 60 phospho-proteins at four time points as a function of 26 perturbations. We perturbed epidermoid carcinoma cells with diverse growth factors and/or small molecules chosen to activate or inhibit specific subsets of receptor tyrosine kinases. The abundance of 60 phosphosites was quantified using a modified microwestern array, a high-confidence assay of protein abundance and modification . Results and Discussion: From the reconstructed network, we identified enhancement of STAT1 activity as a potential strategy to treat EGFR-hyperactive cancers. Quantification of the relationship between drug dosage and cell viability in a panel of triple-negative breast cancer cell lines validated this newly proposed therapy. Conclusions: The accurate and scalable DIONESUS algorithm allows for quick iteration between experimental design and computational modeling by incorporating systems-level behavior in order to accelerate drug target discovery.