Interpretable, similarity-driven multi-view embeddings from high-dimensional biomedical data

TL;DR

SiMLR is a novel algorithm that creates interpretable low-dimensional embeddings from high-dimensional multi-modal biomedical data by leveraging inter-modality relationships, outperforming existing methods in various applications.

Contribution

The paper introduces SiMLR, a new similarity-driven multi-view linear reconstruction method with a unique objective function and regularization, enabling efficient joint reduction of large multi-modal datasets.

Findings

Outperforms related methods in supervised learning tasks

Effective in multi-omics cancer survival prediction

Useful across neuroimaging datasets

Abstract

Similarity-driven multi-view linear reconstruction (SiMLR) is an algorithm that exploits inter-modality relationships to transform large scientific datasets into smaller, more well-powered and interpretable low-dimensional spaces. SiMLR contributes a novel objective function for identifying joint signal, regularization based on sparse matrices representing prior within-modality relationships and an implementation that permits application to joint reduction of large data matrices, each of which may have millions of entries. We demonstrate that SiMLR outperforms closely related methods on supervised learning problems in simulation data, a multi-omics cancer survival prediction dataset and multiple modality neuroimaging datasets. Taken together, this collection of results shows that SiMLR may be applied with default parameters to joint signal estimation from disparate modalities and may yield practically useful results in a variety of application domains.