Sprecher
Beschreibung
Cell-free platforms are multi-component systems comprising Transcription and Translation (TX-TL) machinery liberated from cellular metabolism [1,2]. These systems are composed of three main compartments, cell lysate, Energy Buffer, and DNA template. The nature of multiplicity in cell-free systems bring along a number of variety in terms of TX-TL yield. The underlying cause unfortunately cannot be attributed to a single compartment as the effects are collective and accumulative during TX-TL. However, it is evident in previous studies that cell-free systems are prone to variability [3,4] and therefore, need to be optimized every time one of the compartments is changed. Active learning approaches have been used so far to identify the optimal conditions for cell-free systems [5,6]. Our approach to investigate the best working conditions for high yield protein expression, i.e. high TX-TL yield, is based on investigating Energy Buffer components for specific cell-lysate and DNA template combinations. Our goal is to scan the experimental space of Energy Buffer combinations that will result in maximized TX-TL performance with the help of active learning algorithms. Our approach is based on Energy Buffer combination generation using active learning algorithms. These combinations are then tested for TX-TL yield using a single batch of cell lysate and a fluorescent protein encoding DNA template. The data obtained from each combination is provided to the algorithm allowing it to update its predictions to be tested. This feed-forward loop is conducted multiple times resulting in TX-TL performances, i.e. protein expression yields, almost 12-fold higher than the Energy Buffer combinations provided in the literature.
