Book Description

Rational design and analysis of protein databanks via data-driven algorithms have significantly accelerated drug discovery, in particular, and a wide range of biological research topics, in general, during last decades. A similar approach is gaining momentum in materials research but has garnered limited attention in areas such as the design of soft interfaces formed by solid-binding peptides at solid materials interfaces. The GEMSEC Laboratory (Genetically-Engineered Materials Science and Engineering Center) has been working towards expanding this strategy in materials research via the development of peptide-based bioelectronic interfaces incorporating solid-binding peptides and single layer materials and, thereby, bridge biology to solid-state devices such as graphene field-effect transistors. We are presented with a challenge in peptide-based materials design as, in general, a vast store of relevant data is not available in materials science that is similar to protein databanks that are available in fields such as molecular biology. Thus, there is need for a knowledge-base, but that requires decades of research to draw on. In the present research, this was accounted by utilizing an innovative integration of combinatorial selection of solid-binding peptides, their rational design and bioinformatics based approach to model specific peptide-material interactions. From a data-base of 10s if not hundreds of peptides selected by this approach, the basis of the present method is to generate libraries of materials specific super-peptides that can attach, assemble and perform specific functions on atomically-flat material surfaces. As solid-state systems, single atomic layer materials, such as graphene and those that provide flat surfaces, such as quartz, have been chosen. Using these libraries, peptides that are capable of binding to their counterpart solid material of interest can be identified by performing combinatorial selection based on phage display approach. Typically, 50+ individual peptides are selected from of an original pool of ~1015 variants, which are then classified based on their binding strength using, e.g., fluorescent microscopy. Needleman-Wunsch based similarity analysis and machine learning algorithms are then used to create a scoring matrix capable of identifying robust and weak binders for the particular material amongst millions of random permutations of amino acid sequences in the peptides. The most powerful of these binders are fed into a decision-tree based rational design consisting of selection rules on hydropathicity, iconicity, aromaticity, and polarity of peptides identified to be capable of self-assembly from the previously conduted experiments. This process filters peptides and identifies those that are capable of strongly binding to as well as readily assembling on the atomically flat solid crystals. These model-based designed peptide sequences are then chemically synthesized and subsequently evaluated experimentally in terms of their binding and assembly characteristics using, e.g., atomic force microscopy to validate the success of the predictive model. As the experimental data become available in the assembly of the peptides under specific experimental parameters that are related to the particular chemistry of the sequences, the approach progressively creates a better outcome. Consequently, the model upon each experimental validation is further improvised and provides further knowledge and supply related sequences to the library to advance peptide-guided functional solid-state materials for practical nanotechnology and nanomedicine applications.