WARRENTON, Va., April 6 - Aiming to study protein structures on the genome-wide scale, researchers are spearheading the development of high-throughput technologies for protein purification, crystallization, and analysis, experts said Friday.
“Structural genomics initiatives require as many tools as possible to minimize human intervention,” Ian Wilson of the Scripps Research Institute said at the Second International Structural Genomics Meeting.
The Scripps team is working with collaborators from Syrrx and the Genomics Institute of the Novartis Research Foundation to miniaturize protein production and crystallization. The advantages of miniaturization include a 50- to 100-fold reduction in the amount of protein that must be isolated, so that the crystallization process takes two to four hours rather than a few days. Crystallization from smaller drops can also reduce reagent and storage costs by a factor of 10.
An automated “nanodrop” system allows a large number of trials with crystallization under many different conditions, thus increasing the odds of high-resolution structure determination, Wilson said.
Wilson stressed that the industrial approach involves addressing a number of bottlenecks simultaneously. “There’s no sense having lots of crystal and then finding out we can’t collect the data,” Wilson said.
Tom Terwilliger of the Los Alamos National Laboratory said that automation of crystal structure solution and refinement would make the structural genomics enterprise more practical.
“Crystallography then becomes a tool for a much wider group of researchers, and we get a more efficient use of synchrotron beamlines,” Terwilliger said.
Looking ahead to a time when crystallography data will be accessed by a broader spectrum of scientists, a team including collaborators from the Los Alamos and Lawrence Berkeley National Laboratories, Stanford, Texas A&M, and Cambridge University are collaborating to develop a software suite called the Python-based Environment for Integrated Crystallography, or PHENIX.
PHENIX is intended to provide 3 angstrom resolution automated structure determination in a form usable by novices but with an extendable framework flexible enough for experts. An important feature of the system is the Project History Database, which will include a complete record of everything that has gone into the structure solution.
PHENIX, which will be open source, is expected to be released as a pilot in about three years, and as a complete product in five. In the interim, the team will also be working on enhancements to current structure determination software that will go beyond local optimization, incorporating information from previously solved structures, and taking into account the length of time required for a solution.
Lee Makowski of the Argonne National Laboratory said that in order to develop the right software tools, scientists must look ahead to the comprehensive functional characterization of gene products.
“How are we going to use the huge data sets we are constructing?” Makowski asked. “The form of the databases that we are creating will impact significantly on the ease with which we will be able to use them.”
Makowski continued: “We’re working under the assumption that looking at the whole universe of protein structures will tell us things we can’t learn from a single structure. We need efficient means to archive, organize, analyze and understand the huge data sets produced, and computational and modeling methods to apply these massive data sets to a comprehensive understanding of biomolecular systems.”