SAF: Smart Aggregation Framework for Revealing Atoms Importance Rank and Improving Prediction Rates in Drug Discovery

TL;DR

This paper introduces a new method for improving drug discovery predictions by better understanding the importance of atoms in molecules.

Contribution

A novel weighted aggregation method using the Boltzmann distribution to improve graph neural networks in drug discovery.

Findings

The proposed aggregation method improves prediction of antibiotic activity using graph neural networks.

The method reveals important atoms for activity prediction by adjusting a temperature hyperparameter.

The approach successfully identifies functional groups in β-lactam antibiotics.

Abstract

Machine learning,
and representation learning in particular, has
the potential to facilitate drug discovery by screening a large chemical
space in silico. A successful approach for representing molecules
is to treat them as graphs and utilize graph neural networks. One
of the key limitations of such methods is the necessity to represent
compounds with different numbers of atoms, which requires aggregating
the atom’s information. Common aggregation operators, such
as averaging, result in a loss of information at the atom level. In
this work, we propose a novel aggregating approach where each atom
is weighted nonlinearly using the Boltzmann distribution with a hyperparameter
analogous to temperature. We show that using this weighted aggregation
improves the ability of the gold standard message-passing neural network
to predict antibiotic activity. Moreover, by changing the temperature
hyperparameter, our approach can reveal the atoms that are important
for activity prediction in a smooth and consistent way, thus providing
a novel regulated attention mechanism for graph neural networks. We
further validate our method by showing that it recapitulates the functional
group in β-lactam antibiotics. The ability of our approach to
rank the atoms’ importance for a desired function can be used
within any graph neural network to provide interpretability of the
results and predictions at the node level.

Figures