Home | About IPCCSB | Core Research Mission

  About Intel® PCCSB

Core Research Mission

Understanding the structure-function relationship of biological macromolecules represents a central focus common to much molecular biomedical research in the contemporary life sciences. Among the technologies available for biological structure analysis, cryo-electron microscopy (cryo-EM) is evolving to become a primary tool to visualize the three-dimensional (3D) structures of single biomolecules in their native functional states, potentially at the systems level. However, because biomolecules are highly sensitive to radiation damage by the electron beam, the molecular images have to be taken at a low dose that gives rise to an extremely high degree of noise in the formation of the image. This situation leads to one of the most critical challenges facing computational approaches to cryo-EM analysis of biomolecules; namely, the extraction of signal from heavy noise. This involves the analysis of a large number of very noisy images that allows one to reconstruct the entire structure of the molecule up to atomic resolution through averaging and statistical techniques. Such a procedure is highly data-intensive and computationally demanding; the computing cost increases dramatically with increases in resolution and structural diversity, or with a decrease of the signal-to-noise (SNR) ratio.

Instead of simply migrating the existing cryo-EM software codes, the research at the Intel® PCCSB aims to build a coherent software-hardware system that implements innovative machine-learning methods for massively parallel cryo-EM data processing. The particular focus of these efforts will be weak signal extraction, 3D reconstruction and verification at the single-molecule and systems level, taking full advantage of Intel MIC architecture in a supercomputing environment. We anticipate that the system will become the first of its kind in computational tools for structural biology, being able to process a huge amount of highly noisy image data for structure determination in a heretofore unachievable manner. Further developments along this avenue will produce a new generation of supercomputing platforms for ultra-high-resolution reconstruction of single biomolecules in their native states. Such platforms may emerge as a general resource for future parallel computing applications in structural biology and molecular medicine.