Search for: All records

Abstract Historically, humans have cleared many forests for agriculture. While this substantially reduced ecosystem carbon storage, the impacts of these land cover changes on terrestrial gross primary productivity (GPP) have not been adequately resolved yet. Here, we combine high-resolution datasets of satellite-derived GPP and environmental predictor variables to estimate the potential GPP of forests, grasslands, and croplands around the globe. With a mean GPP of 2.0 kg C m −2  yr −1 forests represent the most productive land cover on two thirds of the total area suitable for any of these land cover types, while grasslands and croplands on average reach 1.5 and 1.8 kg C m −2  yr −1 , respectively. Combining our potential GPP maps with a historical land-use reconstruction indicates a 4.4% reduction in global GPP from agricultural expansion. This land-use-induced GPP reduction is amplified in some future scenarios as a result of ongoing deforestation (e.g., the large-scale bioenergy scenario SSP4-3.4) but partly reversed in other scenarios (e.g., the sustainability scenario SSP1-1.9) due to agricultural abandonment. Comparing our results to simulations from state-of-the-art Earth System Models, we find that all investigated models deviate substantially from our estimates and from each other. Our maps could be used as a benchmark to reduce this inconsistency, thereby improving projections of land-based climate mitigation potentials. 

Protein complex formation is a central problem in biology, being involved in most of the cell's processes, and essential for applications, e.g. drug design or protein engineering. We tackle rigid body protein-protein docking, i.e., computationally predicting the 3D structure of a protein-protein complex from the individual unbound structures, assuming no conformational change within the proteins happens during binding. We design a novel pairwise-independent SE(3)-equivariant graph matching network to predict the rotation and translation to place one of the proteins at the right docked position relative to the second protein. We mathematically guarantee a basic principle: the predicted complex is always identical regardless of the initial locations and orientations of the two structures. Our model, named EquiDock, approximates the binding pockets and predicts the docking poses using keypoint matching and alignment, achieved through optimal transport and a differentiable Kabsch algorithm. Empirically, we achieve significant running time improvements and often outperform existing docking software despite not relying on heavy candidate sampling, structure refinement, or templates.