Coping with complexity: machine learning optimization of cell-free protein synthesis.

Biological systems contain complex metabolic pathways with many nonlinearities and synergies that make them difficult to predict from first principles. Protein synthesis is a canonical example of such a pathway. Here we show how cell-free protein synthesis may be improved through a series of iterated high-throughput experiments guided by a machine-learning algorithm implementing a form of evolutionary design of experiments (Evo-DoE).

Automated discovery of novel drug formulations using predictive iterated high throughput experimentation.

Background: We consider the problem of optimizing a liposomal drug formulation: a complex chemical system with many components (e.g., elements of a lipid library) that interact nonlinearly and synergistically in ways that cannot be predicted from first principles.

Doubly stochastic Poisson distribution of platelet adhesion on material surfaces and its implication on fluorescence image analysis.

An image based assay has been developed to quantify platelet adhesion on material surfaces. Briefly, citrated platelet rich plasma (PRP) is incubated with materials for 2 h to allow platelet adhesion on the surface, followed by fluorescence labeling of platelets with Celltracker Green. Multiple images are acquired by an automatic fluorescence microscope, IN Cell Analyzer 1000.

Development of combinatorial chemistry methods for coatings: high-throughput weathering evaluation and scale-up of combinatorial leads.

Combinatorial screening of materials formulations followed by the scale-up of combinatorial leads has been applied for the development of high-performance coating materials for automotive applications. We replaced labor-intensive coating formulation, testing, and measurement with a "combinatorial factory" that includes robotic formulation of coatings, their deposition as 48 coatings on a 9x12-cm plastic substrate, accelerated performance testing, and automated spectroscopic and image analysis of resulting performance. This high-throughput (HT) performance testing and measurement of the resulting properties provided a powerful set of tools for the 10-fold accelerated discovery of these coating materials.

Efficient discovery of nonlinear dependencies in a combinatorial catalyst data set.

Exploration of a complex catalyst system using Genetic Algorithm methods and combinatorial experimentation efficiently removes noncontributing elements and generates data that can be used to model the remaining system. In particular the combined methods effectively navigate and optimize systems with highly nonlinear dependencies (3-way and higher interactions).

Development of combinatorial chemistry methods for coatings: high-throughput adhesion evaluation and scale-up of combinatorial leads.

Coupling of combinatorial chemistry methods with high-throughput (HT) performance testing and measurements of resulting properties has provided a powerful set of tools for the 10-fold accelerated discovery of new high-performance coating materials for automotive applications. Our approach replaces labor-intensive steps with automated systems for evaluation of adhesion of 8 x 6 arrays of coating elements that are discretely deposited on a single 9 x 12 cm plastic substrate. Performance of coatings is evaluated with respect to their resistance to adhesion loss, because this parameter is one of the primary considerations in end-use automotive applications.