When picturing a protein, a long folded strand might come to mind. That image of a complex structure delicately folding upon itself is possible in part to the advances in specialized computer image modeling software like AlphaFold, which now is available to Indiana University researchers following a major upgrade to Slate-Scratch.
In Spring 2022, the AlphaFold DB (2.3 TB) was placed on dedicated flash storage in order to provide users with better performance compared to spinning disk by Research Technologies. Now, AlphaFold is 3.83 times faster when accessed through its new home on Slate-Scratch.
Distinguished researchers and professors of Biology and Chemistry are weighing in on how access to AlphaFold has improved their time to science, and alleviated computer requirements. By accessing the software via HPC researchers no longer need to download and maintain their own copies of the AlphaFold.
Roger Innes, in the Department of Biology, uses AlphaFold to model how proteins respond to viral and bacterial pathogens in plants.
“By understanding the structure of these intracellular receptors and how they activate immune responses, we can make modifications that will enable plants to detect a broader range of pathogens, and thus make plants resistant to infection without employing chemical pesticides, reducing the environmental impact of agriculture and reducing costs to farmers.” Innes said. “The AlphaFold program has enabled us to rapidly determine the structure of numerous plant immune receptors and pathogen proteins, greatly accelerating our research.”
Ankur Dalia’s Biology lab also studies pathogenic bacteria, but those which directly affect humans. AlphaFold is employed to run models of Vibrio cholerae, the water borne bacteria which causes Cholera. This bacteria often interact with the shells of crustacean zooplankton, which are made of chitin. “One critical factor that is required for V. cholerae to sense and respond to chitin in its environment is the membrane-embedded transcriptional regulator ChiS. ChiS cannot directly sense chitin, but instead relies on a partner protein called CBP, which directly binds to chitin,” Dalia explained.
By looking to interrupt this attraction at a protein level, the lab hopes to disrupt the pathogen in the environment, thus stopping the chain of transmission to humans. “The exact mechanism by which CBP and ChiS interact, and how chitin-binding to CBP helps induce ChiS activity remains a mystery. Thus, my lab has recently employed AlphaFold multimer on Carbonate to generate a structural model of the ChiS-CBP interaction interface. We are now generating mutant alleles of CBP and ChiS in the lab to test this structural model.”
Yasmine Zubi, a graduate student in Chemistry, says the program’s ability to “scratch” out unworkable models by running multiple simulations, helps her lab work on protein engineering when working with enzyme structures that have not yet been published.
“When improving selectivity for a reaction involving a protein with an unknown structure and a particular substrate, we can first use AlphaFold to generate a model of the protein, and then use docking simulations to predict the substrate’s binding site and select sites for targeting,” Zubi said. The GPUs on IU’s high-performance compute systems in conjunction with the Slate file system makes it possible to run several models at once: “AlphaFold enables us to do this in high throughput and generate dozens of structures in parallel,” she said.
Finally, AlphaFold is used as a way to decode existing bio banks of complex information in Irene Newton’s Biology lab. Focusing on the function on novel proteins in bacterial symbiont genomes, “AlphaFold, by predicting the structure of the proteins, will give us a glimpse into potential function using structural homology with proteins in the existing PDB databank,” Newton explained, referring to the RCSB Protein Data Bank.
Learn more about running AlphaFold for your research with RT Projects.
