Intrinsic-Extrinsic Convolution and Pooling for Learning on 3D Protein Structures

Pedro Hermosilla

Ulm University

Marco Schäfer

University of Tübingen

Matej Lang

Masaryk University (Brno)

Gloria Fackelmann

Ulm University

Pere-Pau Vázquez

Universitat Politècnica de Catalunya

Barbora Kozlíková

Masaryk University (Brno)

Michael Krone

University of Tübingen

Tobias Ritschel

University College London

Timo Ropinski

Ulm University

International Conference on Learning Representations 2021

Abstract

Proteins perform a large variety of functions in living organisms and thus play a key role in biology. However, commonly used algorithms in protein learning were not specifically designed for protein data, and are therefore not able to capture all relevant structural levels of a protein during learning. To fill this gap, we propose two new learning operators, specifically designed to process protein structures. First, we introduce a novel convolution operator that considers the primary, secondary, and tertiary structure of a protein by using n-D convolutions defined on both the Euclidean distance, as well as multiple geodesic distances between the atoms in a multi-graph. Second, we introduce a set of hierarchical pooling operators that enable multi-scale protein analysis. We further evaluate the accuracy of our algorithms on common downstream tasks, where we outperform state-of-the-art protein learning algorithms.

Bibtex

@inproceedings{hermosilla2021proteins,
	title={Intrinsic-Extrinsic Convolution and Pooling for Learning on 3D Protein Structures},
	author={Hermosilla, Pedro and Sch{\"a}fer, Marco and Lang, Matej and Fackelmann, Gloria and V{\'a}zquez, Pere-Pau and Kozl{\'i}kov{\'a}, Barbora and Krone, Michael and Ritschel, Tobias and Ropinski, Timo},
	booktitle={Proceedings of International Conference on Learning Representations}
	year={2021}
}