RNA’s role has long been thought to be primarily one of a genetic messenger, while protein does most of the heavy lifting in cells. But some RNAs fold into complex shapes, allowing them to drive cellular processes and catalyze biochemical reactions. Scientists are just beginning to understand the many functions of these active molecules—called noncoding RNAs—in signaling, cancer, development, stress response, and more. Building accurate physical and chemical models of these noncoding RNAs will be critical for determining how they can be harnessed for medicine and other applications.

Molecular Biophysics and Integrated Bioimaging (MBIB) scientists, in collaboration with researchers from the Hebrew University of Jerusalem, have developed a streamlined pipeline for more reliably predicting and validating the structure of an RNA molecule at the atomic level. MBIB staff scientist Michal Hammel and computational research scientist Scott Classen co-developed the process, called SOlution Conformation PrEdictor for RNA (SCOPER), with collaborators Dina Schneidman-Duhovny and Edan Patt.

Creating SCOPER required the team to overcome longstanding roadblocks in characterizing the structure of RNA molecules. While methods like X-ray crystallography can accurately produce three-dimensional atomic blueprints of proteins and AI programs can predict protein structures from their amino acid sequences alone, several challenges remain in applying these technologies to RNAs. RNAs rarely convert into the neat crystalline forms required for X-ray crystallography imaging. And because the twists and folds of the RNA strand move around as the molecule functions, there are actually multiple correct structures. Furthermore, nuanced atomic factors, such as charge and the presence of ions like magnesium, affect how RNAs fold. Composing a picture of a given RNA molecule currently involves the arduous process of combining outputs from multiple computational tools with imaging data, and final results still vary in their accuracy.

“Programs like AlphaFold will sometimes come up with five different models of an RNA. And now the question is, which one is right?” said Hammel. “SCOPER can tell you.” 

To develop SCOPER, the researchers leveraged an existing AI tool that accounts for RNA’s flexibility and developed a new program that predicts the presence of magnesium ions, then combined them into one software pipeline. Members of the research community can come to Berkeley Lab’s Advanced Light Source (ALS) user facility knowing nothing more than a RNA’s nucleotide sequence, and walk away with a set of precise, three-dimensional atomistic models.

While the pipeline can currently be run on the computers at the SYBILS beamline, the team is tapping into supercomputing resources at Berkeley Lab’s National Energy Research Scientific Computing Center (NERSC) user facility to make SCOPER easier and faster to use. Once finished, Berkeley Lab will be a one-stop shop for visualizing the solution state of RNAs: researchers will be able to send in samples by mail, use the automated tools at SIBYLS, and run SCOPER online—no need to visit in person or handle complex software. The team also plans to offer SCOPER as a free, web-based tool with source code, so scientists everywhere can use it without needing to install or set up anything themselves. 

Read more in the Berkeley Lab press release.