(173b) Designing Chimeric Peptides Using Swarm Intelligence Optimization for the Protein Folding Problem

Authors: 
Boone, K., University of Kansas
Tamerler, C., University of Kansas
Camarda, K. V., University of Kansas

Chimeric peptides offer biological engineering methodologies adapted in materials science for wide range of applications including assembly and synthesis.  Tailored chimeric functionality depends not only the sequence but also the molecular structure of the resulting molecules. Designing new chimeric peptides therefore relies on developing a better understanding of the protein folding.   Here, a new ab initio method for finding peptide backbone angles was developed with a swarm intelligence algorithm to find peptide folds faster than using a molecular dynamics approach.  Currently, many approaches to the protein folding problem have been investigated.  The first approach is to experimentally determine the structures.  This is the most accurate way of determining the structure, but the process is slow and requires commitments of large resources.  Determining a structure experimentally without knowing that the peptide design is already successful is prohibitive.  The second approach is to use molecular dynamics simulations.  The main limitation of these simulations for peptide design is the low number of peptides that can be analyzed.  Having larger peptide design candidate pools should help to improve chimeric peptide design.  Using existing data avoids the cost of determining new structures and statistical analysis avoids the computational cost of simulating the protein folding process in the second approach.  However in swarm intelligence algorithm, statistical analysis of determined folding structures may avoid the challenges of the first two approaches.  In previous work1, an antimicrobial peptide that self-assembles onto titanium implants has been developed.  Here, we applied two optimization strategies to these peptide sets for finding chimeric peptide folds more effectively. We next compared their folding patterns to engineer sequence-structure function relations.  The first optimization strategy uses a swarm intelligence algorithm inspired by bee hive coomunication to coordinate the selection of highly likely folds.  The second approach is to match the peptide fragments with historical data of protein folds2.  PyRosetta software3 was used to score the likelihood of structures and to find the bond angles for the amino acid side chains. The likelihood of the structures for each approach was compared, and a time analysis was performed. Our approach offers to expand the pool of new chimeric peptide design. The NIH-NIAMS (7R21AR062249-03) is acknowledged.

 

[1] Yucesoy D, Hnilova M, Boone K, Arnold P, Snead M, Tamerler C. Chimeric Peptides as Implant Functionalization Agents for Titanium Alloy Implants with Antimicrobial Properties. JOM 2015;67:754-66.

[2] Kim DE, Chivian D, Baker D. Protein structure prediction and analysis using the Robetta server. Nucleic Acids Research 2004;32:W526-W31.

[3] Chaudhury S, Lyskov S, Gray JJ. PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta. Bioinformatics 2010;26:689-91.