A Computational Model for Precisely Decoding Complex Chemical Mixtures with Chemosensory Arrays

Web Published:

Princeton Docket # 10-2599

To date, analysis of chemosensory array (¿artificial nose¿) data has been generally limited to ¿fingerprinting¿ or ¿pattern recognition¿ methods of data analysis. These methods do not factor in the physical nature of mixed, cross-inhibitory signals from multiple ligands. This means that chemosensory arrays based on non-specific receptors are not optimized to account for ¿crosstalk¿ among analytes, which significantly limits their utility. It also means that sensory arrays must frequently rely on highly specific ligand-receptor interactions if they are to be specific for a given ligand, which limits the number of ligands recognized by the array.


Researchers in the Departments of Molecular Biology, Princeton University and the Department of Physics and Astronomy, Rutgers University have developed a method for statistical analysis of chemosensory array data based on a physical model that accounts for cross-inhibitory signals in mixtures of ligands.  This model facilitates analysis of sensory data from chemical sensory arrays, making it easier to design new assays, deconvolute chemical ¿signatures¿, make quantitative predictions of the concentration of each component in complex mixtures, and validate results.  


The computational algorithm utilizes a Bayesian inference technique, applied to a pharmacological model for receptor-analyte interactions.  Sensory data is independently compiled for each receptor-analyte pair; sensory data from unknown mixtures can then be interpreted in terms of the known response of the sensory array system to individual ligands.  This methodology can be employed with both biological systems and artificial receptor arrays (¿electronic noses¿) designed for various industrial needs. Currently, the researchers are developing arrays of electrochemical sensors for real-time detection of harmful gases such as NO, NO2, and CO in complex mixtures such as diesel engine exhaust.



·         Diagnostics

·         Chem/bio weapons detection

·         Drug tests

·         Food monitoring

·         Environmental monitoring

·         Quality/process control



·         Accounts for combinatorial mixtures of chemical analytes

·         Quantitative or semi-quantitative

·         Lower error rates

·         Better quality control

·         Better discriminatory power



Tsitron J, Ault AD, Broach JR, Morozov AV. ¿Decoding complex chemical mixtures with a physical model of a sensor array.¿  PLoS Comput Biol. 2011 Oct;7(10):e1002224.


The Inventors

James Broach is Professor and Associate Chair of Department of Molecular Biology. The research in Professor Broach¿s laboratory is directed toward understanding cellular regulation at the molecular level, using the yeast Saccharomyces cerevisiae as a model system.  Dr. Broach was Co-Founder and Director of Research of Cadus Pharmaceuticals Corporation (Cadus), a biotechnology firm with drug discovery programs that focused on G-protein coupled receptors and utilized a core yeast technology for developing drug discovery assays. Dr. Broach also serves as a member of the Science Advisory Board for the US Food and Drug Administration.

Addison Ault is post-doctoral fellow in Prof. Broach¿s laboratory.

Alexandre V. Morozov is Assistant Professor in the Department of Physics & Astronomy at Rutgers University.  Professor Morozov¿s research is focused on applications of statistical physics to biology, including prediction of gene regulation on a whole-genome scale and its variation influenced by cell type, environmental signals, developmental stage, and disease state.

Julia Tsitron is a Ph.D. student in Professor Morozov¿s laboratory.


Intellectual Property status

Patent protection is pending.

Patent Information:
For Information, Contact:
Laurie Tzodikov
Licensing Associates
Princeton University
James Broach
Addison Ault
Alex Morozov
Julia Tsitron
medical device
platform technology