Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:

Supercomputing Plant Polymers for Biofuels

plant-biofuel_simA huge barrier in converting cellulose polymers to biofuel lies in removing other biomass polymers that subvert this chemical process. To overcome this hurdle, large-scale computational simulations are picking apart lignin, one of those inhibiting polymers, and its interactions with cellulose and other plant components. The results point toward ways to optimize biofuel production and are helping researchers understand the complex chemistry of plant cell walls.

Corn-based ethanol has been a popular but controversial source of renewable energy. Producing ethanol from corn is energy- intensive, yielding just 25 percent more than what was required to produce it, said a 2006 study published in Proceedings of the National Academies of Sciences. In addition, corn used to produce energy can’t be used as food.

Converting cellulose into ethanol could be a more efficient way to produce biofuels. But to unlock the energy in this carbohydrate polymer found in all plants, bioengineers must first separate it from lignin and other plant biomass components. The polymers of choice can then be converted into other useful chemicals – for example, simple sugars that microbes can digest and convert to ethanol.

Little is known about the structure and chemical behavior of lignin – which, for instance, helps provide wood’s solidity and structure – and its detailed makeup varies depending on the plant species. Lignin’s complexity and the way it’s entwined with cellulose have made it a candidate for large-scale computational simulation.

Researchers at the Center for Molecular Biophysics at Oak Ridge National Laboratory (ORNL) and the University of Tennessee (UT) in Knoxville have long used supercomputers to model and study lignin polymers and their interactions with cellulose. More recently they’ve started to include other biomass polymers, such as hemicelluloses and pectin, with the idea of simulating all the chemical components of plant cell walls. They’re now applying their second INCITE allotment to the task: a hundred-million processor hours on Titan, ORNL’s Cray XK7 supercomputer.

Plants have evolved to avoid degradation by bacteria or fungi, says Jeremy Smith of ORNL and UT. Though biomass offers a rich source of energy and chemicals, researchers have to overcome these plant survival strategies. Scientists are exploring a few possible solutions: Plants can be engineered so they’re more vulnerable to hydrolases, enzymes that break down cellulose into simple sugars, and microbes can be engineered to break down cellulose more efficiently.

Smith and his colleagues have been particularly interested in dealing with the previously unwanted gunk such as lignin. It’s a greasy polymer, insoluble in water, and tends to stick both to the cellulose and to the enzymes used to digest it while clumping together. Chemists have used computers for decades to model biomolecules such as proteins and DNA, says Loukas Petridis, an ORNL staff scientist. A technique called molecular dynamics simulation solves Newton’s equation of motion for each individual atom over incredibly small time intervals, Smith says. Those snapshots join into movies that can show molecular behavior over a microsecond. Following millions of atoms over billions of time steps requires supercomputer power.

Petridis, Smith and their colleagues collaborate with experimental researchers at the ORNL-based Spallation Neutron Source and other facilities, helping to ensure that their observations and predictions are consistent with the experimental behavior of biomass. “Nearly everything we do is in collaboration with an experimentalist,” Smith says.

The team is moving on to more complex questions related to biofuel production from cellulose.

They started out by examining the structure and molecular behavior of millions of atoms in lignin – and how it behaves in water and changes with subtle alterations in its sequence and structure. Then they looked at a combination of lignin and cellulose fibers to understand exactly how these polymers stick together.

A critical question was how lignin interacts with hydrolases to convert cellulose to simple sugars for fermentation into ethanol. Bioengineers knew lignin inhibited enzymatic reactions but not how. Using simulations, Petridis, Smith and colleagues showed that lignin binds to the hydrolase enzymes in the exact location where those enzymes bind to cellulose to cleave it apart, effectively blocking the decomposition reaction.

With their most recent INCITE allotment, the team is moving beyond basic understanding of lignin to more complex questions related to biofuel production from cellulose. Earlier this year the researchers collaborated with University of California, Riverside, scientists Charles Wyman and Charles Cai, who were trying to figure out why pretreatment with a common, nontoxic chemical solvent, tetrahydrofuran (THF), made ethanol production from cellulose more efficient.

So the ORNL-UT team went to work with molecular dynamics simulations. The calculations showed that THF inserts itself between lignin and cellulose, Smith says. “It stops the lignin and the cellulose from sticking to each other and stops the lignin from sticking to itself.” The strategy could prove useful as a way to split off lignin, Petridis adds, which could prove useful as DOE researchers look for ways to use other biomass components to produce energy and other useful products.

To expand their simulations from cellulose polymers to lignin to ever more complicated biomolecule combinations, the team has had to optimize communication between parallel processors in supercomputers such as Titan. The machine’s capability should let them scale up to the point that the team can simulate atoms in parts of a plant’s cell wall. The researchers will start by incorporating other cell wall biopolymers such as hemicelluloses and pectin and then add various enzymes.

We can envisage simulations of the complete cell wall,” Smith says. Then they could expand to cell walls from a variety of plant species and even interactions between the plant cell walls and microbial surfaces. With exascale computing power – roughly 100 times that of Titan – Smith expects that the team might eventually simulate the workings of an entire plant cell.

Source: ASCR

Sign up for our insideHPC Newsletter

Leave a Comment


Resource Links: