Petascale computers allow scientists and engineers not only to address old problems better, but also to consider new methods and new problems. We report here on work that both applies new methods and tackles new problems in the area of structural biology. The project combines an efficient protein structure prediction algorithm implemented in the Open Protein System (OOPS) system with the Swift parallel scripting system to enable the rapid and flexible composition of OOPS components into parallel program, and the high-performance execution of these programs on petascale computers. The result is a powerful computational laboratory environment for predicting protein secondary and tertiary structure, for further testing and refining OOPS, and for performing training and scaling tests that enable structure simulation to run on a wide variety of computing architectures with high efficiency. Comparison of before and after experiences within two laboratories at the University of Chicago shows that this use of scripting enables achieving significant improvements in throughput, time-to-solution, and scientific productivity. For example, an undergraduate has recently been able to define and execute new protein folding simulations on thousands of processors. This approach both enables new applications for petascale computers, and provides an avenue for many more researchers to participate in the computational science aspects of structure prediction.