Multitask Protein Function Prediction Through Task Dissimilarity

TL;DR

This paper introduces a multitask learning approach for protein function prediction that leverages task dissimilarity to better handle unbalanced data and improve stability across different evaluation settings.

Contribution

It proposes a novel use of task dissimilarity in multitask learning, demonstrating improved performance over similarity-based methods in protein function prediction.

Findings

Dissimilarity-based multitask learning outperforms similarity-based methods.

The method is more stable across different evaluation settings.

Empirical results on three model organisms support the approach.

Abstract

Automated protein function prediction is a challenging problem with distinctive features, such as the hierarchical organization of protein functions and the scarcity of annotated proteins for most biological functions. We propose a multitask learning algorithm addressing both issues. Unlike standard multitask algorithms, which use task (protein functions) similarity information as a bias to speed up learning, we show that dissimilarity information enforces separation of rare class labels from frequent class labels, and for this reason is better suited for solving unbalanced protein function prediction problems. We support our claim by showing that a multitask extension of the label propagation algorithm empirically works best when the task relatedness information is represented using a dissimilarity matrix as opposed to a similarity matrix. Moreover, the experimental comparison carried out on three model organism shows that our method has a more stable performance in both "protein-centric" and "function-centric" evaluation settings.