Microbiome Immunity Project Researchers Create New Method of Predicting Protein Function

The Microbiome Immunity Project has (so far) identified almost 275,000 unique protein structures, returned more than 430 million work units, and led to the creation of an important new scientific technique.

A growing number of unique protein structures

Thanks to the enthusiastic participation of World Community Grid volunteers, the Microbiome Immunity Project has identified almost 275,000 unique protein structures to-date. You’re continuing to help us generate new results faster than ever!

Our new protein function prediction method

In addition to building and sending work units, we’ve been hard at work for the past six months on creating a new protein function prediction method which uses both a protein’s sequence and its structure. Understanding protein function is important because this knowledge helps scientists explore how the bacterial proteins interact with each other and their hosts, and determine which proteins or biochemical pathways may play a role in any number of diseases.

The new method we’ve developed is more effective than any other technique developed so far because it uses protein structure information, which is crucial to understanding how biology works. State-of-the-art methods to date relied only on protein sequence information, in part because of technological limitations, and because researchers never had access to enough protein structure data to make such predictions widely applicable. We tackled both of those problems; the technological one was addressed by developing a cutting-edge deep learning-based method using graph convolutional neural networks, that provides superior predictive power; and the data availability problem by enrolling your help through the Microbiome Immunity Project and generating a collection of 3D protein structural models of an unprecedented scale.

The most unique and important feature of our method is that it works equally well with experimental 3D structures and computational models, like the ones we generate through the Microbiome Immunity Project. That’s why the two are a perfect match: our state-of-the-art function prediction method and Microbiome Immunity Project results. Significantly, our method is currently the only one that uses 3D structures.

We are so excited about our new function prediction method that we shared our results online as a pre-print, which you can read here. We are working now on a full paper that we plan to submit to a peer-reviewed journal. We’ll let everyone know once it’s published.

Next steps

We’re now planning to annotate all known human gut microbiome data using our new methodology combined with Microbiome Immunity Project results. This will give us higher-quality and higher-coverage data, which will help us, and eventually other researchers, better understand how the microbiome works.

We appreciate your support for this project!


Related Articles