Machine learning to determine optimal conditions for controlling the size of elastin-based particles.

This paper evaluates the aggregation behavior of a potential drug and gene delivery system that combines branched polyethyleneimine (PEI), a positively-charged polyelectrolyte, and elastin-like polypeptide (ELP), a recombinant polymer that exhibits lower critical solution temperature (LCST). The LCST behavior of ELP has been extensively studied, but there are no quantitative ways to control the size of aggregates formed after the phase transition. The aggregate size cannot be maintained when the temperature is lowered below the LCST, unless the system exhibits hysteresis and forms irreversible aggregates. This study shows that conjugation of ELP with PEI preserves the aggregation behavior that occurs above the LCST and achieves precise aggregate radii when the solution conditions of pH (3, 7, 10), polymer concentration (0.1, 0.15, 0.3 mg/mL), and salt concentration (none, 0.2, 1 M) are carefully controlled. K-means cluster analyses showed that salt concentration was the most critical factor controlling the hydrodynamic radius and LCST. Conjugating ELP to PEI allowed crosslinking the aggregates and achieved stable particles that maintained their size below LCST, even after removal of the harsh (high salt or pH) conditions used to create them. Taken together, the ability to control aggregate sizes and use of crosslinking to maintain stability holds excellent potential for use in biological delivery systems.

Evaluation of machine learning algorithms to predict the hydrodynamic radii and transition temperatures of chemo-biologically synthesized copolymers.

Elastin-like polypeptides (ELP) belong to a family of recombinant polymers that shows great promise as biocompatible drug delivery and tissue engineering materials. ELPs aggregate above a characteristic transition temperature (Tt). We have previously shown that the Tt and size of the resulting aggregates can be controlled by changing the ELP's solution environment (polymer concentration, salt concentration, and pH). When coupled to a synthetic polyelectrolyte, polyethyleneimine (PEI), ELP retains its Tt behavior and gains the ability to be crosslinked into defined particle sizes. This paper explores several machine learning models to predict the Tt and hydrodynamic radius (Rh) of ELP and two ELP-PEI polymers in varying solution conditions. An exhaustive design of experiments matrix consisting of 81 conditions of interest with varying salt concentration (0, 0.2, 1 M NaCl), pH (3, 7, 10), polymer concentration (0.1, 0.17, 0.3 mg/mL), and polymer type (ELP, ELP-PEI800, ELP-PEI10K) was investigated. The five models used in this study were multiple linear regression, elastic-net, support vector regression, multi-layer perceptron, and random forest. A multi-layer perceptron model was found to have the highest accuracy, with an R2 score of 0.97 for both Rh and Tt. This was followed closely by the random forest model, with an R2 of 0.94 for Rh and 0.95 for Tt. Feature importance was determined using the random forest and linear regression models. Both models showed that salt concentration and polymer type were the two most influential factors that determined Rh, while salt concentration was the dominant factor for Tt.