Deep structure integrative representation of multi-omics data for cancer subtyping

Abstract

Motivation Cancer is a heterogeneous group of diseases. Cancer subtyping is a crucial and critical step to diagnosis, prognosis and treatment. Since high-throughput sequencing technologies provide an unprecedented opportunity to rapidly collect multi-omics data for the same individuals, an urgent need in current is how to effectively represent and integrate these multi-omics data to achieve clinically meaningful cancer subtyping. Results We propose a novel deep learning model, called Deep Structure Integrative Representation (DSIR), for cancer subtypes dentification by integrating representation and clustering multi-omics data. DSIR simultaneously captures the global structures in sparse subspace and local structures in manifold subspace from multi-omics data and constructs a consensus similarity matrix by utilizing deep neural networks. Extensive tests are performed in 12 different cancers on three levels of omics data from The Cancer Genome Atlas. The results demonstrate that DSIR obtains more significant performances than the state-of-the-art integrative methods. Availability and implementation https://github.com/Polytech-bioinf/Deep-structure-integrative-representation.git Supplementary information Supplementary data are available at Bioinformatics online.

Deeply integrating latent consistent representations in high-noise multi-omics data for cancer subtyping

“…These include the early integration method LRAcluster [ 13 ], the late integration method PINSPlus [ 16 ] and 12 intermediate integration methods. The intermediate integration methods include six traditional methods and six latest deep learning integration methods: SNF [ 21 ], rMKL-LPP [ 17 ], MCCA [ 14 ], MultiNMF [ 15 ], iClusterBayes [ 12 ], NEMO [ 22 ], DCAP [ 29 ], DLSF [ 30 ], DSIR [ 31 ], MRGCN [ 32 ], MOCSS [ 37 ] and DMCL [ 38 ]. Among them, DCAP, DLSF, MOCSS and DMCL are deep learning integration methods designed to solve the noise in heterogeneous omics data.…”

“…DLSF [ 30 ] proposes to integrate the multi-omics data by learning consistent manifolds in the latent sample space for disease subtypes identification. DSIR [ 31 ] simultaneously captures the global structures in sparse subspace and local structures in manifold subspace from multi-omics data and constructs a consensus similarity matrix by utilizing deep neural networks. MRGCN [ 32 ] simultaneously encodes and reconstructs multiple omics expression and similarity relationships into a shared latent embedding space.…”

MOCSS: Multi-omics data clustering and cancer subtyping via shared and specific representation learning

“…Deep learning methods utilize multiple neural networks to train multi-omics data for obtaining latent representations, which are somehow integrated and fed into downstream clustering tasks. 28 , 29 , 30 , 31 , 32 , 33 , 34 It should be noted that the above-mentioned methods may belong to more than one category at the same time.…”