05.06.2013 Views

PNNL-13501 - Pacific Northwest National Laboratory

PNNL-13501 - Pacific Northwest National Laboratory

PNNL-13501 - Pacific Northwest National Laboratory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

HostBuilder: A Tool for the Combinatorial Generation of Host Molecules<br />

Study Control Number: PN00051/1458<br />

Benjamin P. Hay<br />

Host molecules that selectively complex targeted guests have potential broad application to problems currently<br />

challenging DOE including detection of species in groundwater and chemical process streams; the separation and<br />

concentration of radionuclides and other ionic species from nuclear waste, contaminated soils, and groundwater; and the<br />

encapsulation of radionuclides in medical applications. Motivated by the expense of host development in the laboratory,<br />

we are creating software to evaluate large numbers of possible host architectures and rapidly identify the best candidates<br />

prior to synthesis and testing.<br />

Project Description<br />

The purpose of this project is to create software for rapid<br />

discovery of new host architectures. The approach<br />

involves a combinatorial method in which we examine<br />

every possible architecture for a set of donor groups. In<br />

the first year of the project, we developed and coded an<br />

algorithm for connecting two host fragments to form<br />

candidate structures and to rank-order the results with<br />

respect to the best fit to a guest. In addition, we<br />

constructed a comprehensive link library derived from<br />

small alkane and alkene fragments. Tests demonstrate<br />

that the algorithm is computationally very efficient and<br />

with recent hardware we estimate that speeds in excess of<br />

one million host molecules per second will be attainable.<br />

Code development remains in progress and a final version<br />

of the software will be completed in 2001.<br />

Introduction<br />

Many fundamental studies of host-guest interactions have<br />

been performed. Effective computational methods are<br />

available for ranking proposed host structures in terms of<br />

their binding affinities for specific guests on a case-bycase<br />

basis. However, no effective method exists for<br />

computer-aided design of new and improved host<br />

molecules. What is missing is an efficient and systematic<br />

means of generating trial structures. To fill this gap, we<br />

are working to develop a code, HostBuilder, that will be<br />

capable of generating and rank-ordering millions of<br />

possible host architectures from a given set of donor<br />

groups. This new capability will provide DOE with an<br />

alternative to expensive Edisonian experimentation that<br />

currently dominates host-development research.<br />

The objective is to develop a computational approach for<br />

discovering new host molecule structures with the<br />

potential to tightly bind to a specified guest. To achieve<br />

130 FY 2000 <strong>Laboratory</strong> Directed Research and Development Annual Report<br />

this goal, we are combining fundamental knowledge of<br />

guest-donor group interactions with a combinatorial<br />

geometric structure generator to produce and rank a large<br />

library of possible host structures. The project has two<br />

distinct tasks: 1) developing and encoding the algorithms<br />

for the combinatorial geometric structure generator,<br />

HostBuilder; and 2) constructing a large database of<br />

linking groups. Our efforts focused on the first task with<br />

the development of a prototype version of the HostBuilder<br />

program. Our future objective will be the improvement of<br />

the HostBuilder program, completion of a larger linker<br />

database, and validation of the final version of the<br />

software. The successful completion of this project will<br />

yield a powerful software package to guide the<br />

development of molecular and ion recognition agents. To<br />

our knowledge, this capability currently does not exist<br />

elsewhere.<br />

Results and Accomplishments<br />

We successfully completed a prototype version of the<br />

HostBuilder code. This code, which consists of<br />

8,500 lines of standard FORTRAN, takes two host<br />

fragments and builds every possible architecture that can<br />

be made by connecting the fragments to linkage structures<br />

taken from a database. By examining all rotamers and<br />

enantiomers, all possible configurations of each molecular<br />

connectivity were considered during the building process.<br />

After construction, each candidate host architecture was<br />

rank-ordered with respect to guest complementarity based<br />

upon knowledge of the optimal coordination geometry of<br />

each donor group. After all candidates were evaluated,<br />

the best structures were written to a graphics file for<br />

viewing. The current linker database contains molecular<br />

fragments representing all possible connectivities that can<br />

be formed from alkanes and alkenes containing four<br />

carbon atoms or less.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!