Why RNA Structure Prediction Matters
Despite this importance, RNA is still much less understood, than for example protein structure. Experimental ways of determining structure, such as X-Ray crystallography or cryo-electron microscopy can provide highly detailed information about a molecule’s structure but are expensive and difficult to scale.
Recent advancements in Deep Learning and natural language processing, has created a rapidly growing field of research that focuses on computational methods for predicting biological markers using omics (Genome, Transcriptome and Proteome), with two notable publications by Google, AlphaFold for Protein structure prediction and more recently Alpha Genome as a DNA language model.
Our Current Focus
The RNA folding Team at BioMatrix is currently focused on exploring and developing cutting edge and novel methods of RNA structure prediction using emerging deep learning technologies.
As a source of guidance and most importantly data, we are currently using the Ribonanza Challenge hosted by Stanford and Eterna in an effort to build a model for predicting RNA secondary structure.
Our short-term goal is to replicate results from the challenge by training and evaluating our own model, using on the results and findings by of competition winners and the final publication by the Authors of the challenge.
Long Term Goals
This will hopefully teach us a lot about how RNA is structured but also how to build and deploy a Deep Learning applications efficiently and robustly.
Eventually, we are also interested in working on our own research and sharing them with the broader research community in the form of a publication, open-source code or similar.