Illuminating Earth’s microbial diversity and origins from metagenomes with deep learning

Project Information

ERMADA

Grant agreement ID: 838018

DOI

10.3030/838018

Project closed

EC signature date 1 April 2019

Start date 1 August 2019

End date 21 March 2025

Funded under

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Total cost

€ 247 628,16

EU contribution

€ 247 628,16

Coordinated by

EREVNITIKO KENTRO VIOIATRIKON EPISTIMON ALEXANDROS FLEMINGK
Greece

Periodic Reporting for period 1 - ERMADA (Illuminating Earth’s microbial diversity and origins from metagenomes with deep learning)

Reporting period: 2019-08-01 to 2022-07-31

Summary of the context and overall objectives of the project

The project coincided with the onset of the worldwide crisis due to the COVID-19 pandemic. As a consequence, the scope and direction of the project were modified in order to combat this unknown threat and address the unprecedented health situation with the latest scientific tools.
Resources and expertise were directed towards an international collaboration of researchers and groups from France, Greece, Canada, USA, Spain and others, with the aim of uncovering the mechanics of the viral invasion in human cells with a combination of state-of-the-art biological and computational tools.
The benefits of understanding the underlying molecular mechanisms of COVID-19 pathogenesis held an immense potential for saving potentially millions of lives and halting the virus' devastating impact on all human activity during the critical first months of the pandemic spread.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The approach followed was a combination of proximal proteomics (the mapping of all the interactions between the viral and human proteins inside the cell volume) with computational techniques for the analysis and visualization of networks in three dimensions.
The large-scale proximal proteomic data lend themselves naturally to modeling by graphs, which when laid out in 3D provide an intuitive and physical picture of the complex processes involved during viral invasion in the cell. This computational approach proved remarkably well-suited by establishing known facts about human proteins' cellular localizations and revealing or corroborating viral mechanisms of action. The results held true when integrating with larger datasets from other proteomics studies employing a variety of different methods. Even more importantly, the 3D modeling of the proximal interactome was able to provide quantitative predictions for direct interactions observed in independent experiments performed as part of the same collaboration. Finally, a select set of protein interactions was also modeled in simulation with cutting edge deep learning tools (AlphaFold2), in order to provide further support of the findings and indicate possible targets for directed therapies.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The 3D modeling and representation of proximal proteomic data was performed for the first time to such scale and depth. The web-based interactive dashboard allowed researchers to explore the wealth of data with numerous customizations and to the most granular level of detail.
Beyond the SARS-CoV-2 virus, the methodology provided a template for quickly investigating an unknown or emerging viral threat and provides a pipeline that could be used in future situations with the aim devising countermeasures or targeted therapies.

Snapshot of 3D network modeling of SARS-CoV-2 cellular invasion