Technology for biocomputational devices based on molecular motors

Sammanfattning: For many multivariable problems there are no efficient algorithms for finding solutions. To this day, conventional electronic computers mainly solve problems in a sequential manner. This sequential operation prevents problem-solving within a reasonable time-frame due to superpolynomial, and sometimes even exponential, time complexities. Developing powerful parallel computation techniques has therefore gained increasing attention in science and industry to overcome this fundamental limitation.This thesis is aimed towards developing network-based biocomputation using molecular motors for solving combinatorial problems in a massively parallel manner. Molecular motors have been previously used to compute a small scale subset sum problem encoded into a network of channels and junctions1. Here, we tackle some of the engineering requirements of upscaling this system. Specifically, we have developed; reliable surface treatments for chemical modification, the regeneration of surfaces for a more sustainable fabrication process, the development of high throughput fabrication with nanometre scale resolution, the design optimisation of the graphical encoding and the development of two architectural elements: programmable gates for versatile networks, and electric sensors for label free detection of filaments.We show how the material chemistry of molecular motor devices can be altered in a controlled way to ensure selective protein binding, only promoting motility in designated areas. To reuse these devices, we developed a method to regenerate the surfaces with a non-destructive approach, which prolongs their life-time and enables them to be used multiple times. We also present a new device system using two new polymer resists with tuneable motility properties.To be able to fabricate large-scale, high-resolution devices within a reasonable time-frame we optimised the patterning parameters for electron beam lithography. We also show that nanoimprint lithography can be used as a high-throughput, high-resolution fabrication method to pattern the type of structures needed for molecular motor devices. By adjusting the structural design and imprinting parameters we are able to fabricate high aspect ratio patterns that successfully promote motility.We also demonstrate a method of translating the exact cover problem into the subset sum problem. We present the design optimisation of a large-scale network (~1000 solutions) encoding the two combinatorial problems mentioned, and our progress towards finding a solution using the molecular motor system actin-myosin II as exploratory agents.Additionally, we describe a method of creating switchable motility to create programmable network-junctions to be able to compute different mathematical encodings. Such dynamic encodings are a necessity for any viable computer. Finally, we present the advancements towards creating an electric sensor for detection of cytoskeletal filaments, using a carbon nanotube as a tripwire, to enable a reliable readout method for highly parallel problem solving.We conclude with a discussion of future challenges and prospects for network-based biocomputation with molecular motors in the light of our findings presented here.