{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T20:33:42Z","timestamp":1772138022559,"version":"3.50.1"},"reference-count":47,"publisher":"Oxford University Press (OUP)","issue":"1","license":[{"start":{"date-parts":[[2023,11,29]],"date-time":"2023-11-29T00:00:00Z","timestamp":1701216000000},"content-version":"vor","delay-in-days":7,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Convergence Health and Technology program"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2023,11,22]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Multi-omic analyses are necessary to understand the complex biological processes taking place at the tissue and cell level, but also to make reliable predictions about, for example, disease outcome. Several linear methods exist that create a joint embedding using paired information per sample, but recently there has been a rise in the popularity of neural architectures that embed paired -omics into the same non-linear manifold. This work describes a head-to-head comparison of linear and non-linear joint embedding methods using both bulk and single-cell multi-modal datasets. We found that non-linear methods have a clear advantage with respect to linear ones for missing modality imputation. Performance comparisons in the downstream tasks of survival analysis for bulk tumor data and cell type classification for single-cell data lead to the following insights: First, concatenating the principal components of each modality is a competitive baseline and hard to beat if all modalities are available at test time. However, if we only have one modality available at test time, training a predictive model on the joint space of that modality can lead to performance improvements with respect to just using the unimodal principal components. Second, -omic profiles imputed by neural joint embedding methods are realistic enough to be used by a classifier trained on real data with limited performance drops. Taken together, our comparisons give hints to which joint embedding to use for which downstream task. Overall, product-of-experts performed well in most tasks and was reasonably fast, while early integration (concatenation) of modalities did quite poorly.<\/jats:p>","DOI":"10.1093\/bib\/bbad416","type":"journal-article","created":{"date-parts":[[2023,10,31]],"date-time":"2023-10-31T13:40:17Z","timestamp":1698759617000},"source":"Crossref","is-referenced-by-count":9,"title":["An in-depth comparison of linear and non-linear joint embedding methods for bulk and single-cell multi-omics"],"prefix":"10.1093","volume":"25","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-3111-4268","authenticated-orcid":false,"given":"Stavros","family":"Makrodimitris","sequence":"first","affiliation":[{"name":"Delft University of Technology Delft Bioinformatics Lab, , Street, Postcode, State, Country"},{"name":"Erasmus University Medical Center Department of Medical Oncology, , Street, Postcode, State, Country"},{"name":"Erasmus University Medical Center Department of Clinical Genetics, , Street, Postcode, State, Country"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0001-8017-1066","authenticated-orcid":false,"given":"Bram","family":"Pronk","sequence":"additional","affiliation":[{"name":"Delft University of Technology Delft Bioinformatics Lab, , Street, Postcode, State, Country"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2719-5736","authenticated-orcid":false,"given":"Tamim","family":"Abdelaal","sequence":"additional","affiliation":[{"name":"Delft University of Technology Delft Bioinformatics Lab, , Street, Postcode, State, Country"},{"name":"Leiden University Medical Center Department of Radiology, , Street, Postcode, State, Country"},{"name":"Leiden University Medical Center Leiden Computational Biology Center, , Street, Postcode, State, Country"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1148-1562","authenticated-orcid":false,"given":"Marcel","family":"Reinders","sequence":"additional","affiliation":[{"name":"Delft University of Technology Delft Bioinformatics Lab, , Street, Postcode, State, Country"},{"name":"Leiden University Medical Center Leiden Computational Biology Center, , Street, Postcode, State, Country"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"286","published-online":{"date-parts":[[2023,11,28]]},"reference":[{"key":"2023112916385421400_ref1","doi-asserted-by":"crossref","DOI":"10.3389\/fgene.2020.610798","article-title":"State of the field in multi-omics research: from computational needs to data mining and sharing","volume":"11","author":"Krassowski","year":"2020","journal-title":"Front Genet"},{"issue":"12","key":"2023112916385421400_ref2","doi-asserted-by":"crossref","first-page":"774","DOI":"10.15252\/msb.20145487","article-title":"Integrated systems analysis reveals a molecular network underlying autism spectrum disorders","volume":"10","author":"Li","year":"2014","journal-title":"Mol Syst Biol"},{"issue":"10","key":"2023112916385421400_ref3","doi-asserted-by":"crossref","first-page":"1141","DOI":"10.1038\/ng.2734","article-title":"The integrated landscape of driver genomic alterations in glioblastoma","volume":"45","author":"Frattini","year":"2013","journal-title":"Nat Genet"},{"issue":"1","key":"2023112916385421400_ref4","doi-asserted-by":"crossref","first-page":"76","DOI":"10.1186\/s13059-021-02293-3","article-title":"Single cell eQTL analysis identifies cell type-specific genetic control of gene expression in fibroblasts and reprogrammed induced pluripotent stem cells","volume":"22","author":"Neavin","year":"2021","journal-title":"Genome Biol"},{"issue":"10","key":"2023112916385421400_ref5","doi-asserted-by":"crossref","first-page":"1113","DOI":"10.1038\/ng.2764","article-title":"The cancer genome atlas pan-cancer analysis project","volume":"45","author":"Weinstein","year":"2013","journal-title":"Nat Genet"},{"issue":"8","key":"2023112916385421400_ref6","doi-asserted-by":"crossref","first-page":"4004","DOI":"10.1038\/s41596-021-00571-9","article-title":"Smart-RRBS for single-cell methylome and transcriptome analysis","volume":"16","author":"Gu","year":"2021","journal-title":"Nat Protoc"},{"issue":"13","key":"2023112916385421400_ref7","doi-asserted-by":"crossref","first-page":"3573","DOI":"10.1016\/j.cell.2021.04.048","article-title":"Integrated analysis of multimodal single-cell data","volume":"184","author":"Hao","year":"2021","journal-title":"Cell"},{"issue":"8","key":"2023112916385421400_ref8","doi-asserted-by":"crossref","first-page":"831","DOI":"10.1016\/j.tig.2022.03.015","article-title":"Into the multiverse: advances in single-cell multiomic profiling","volume":"38","author":"Ogbeide","year":"2022","journal-title":"Trends Genet"},{"key":"2023112916385421400_ref9","doi-asserted-by":"crossref","first-page":"01","DOI":"10.1038\/s41467-020-20430-7","article-title":"Benchmarking joint multi-omics dimensionality reduction approaches for the study of cancer","volume":"12","author":"Cantini","year":"2021","journal-title":"Nat Commun"},{"issue":"111","key":"2023112916385421400_ref10","doi-asserted-by":"crossref","first-page":"111","DOI":"10.1186\/s13059-020-02015-1","article-title":"MOFA+: a statistical framework for comprehensive integration of multi-modal single-cell data","volume":"21","author":"Argelaguet","year":"2020","journal-title":"Genome Biol"},{"key":"2023112916385421400_ref11","doi-asserted-by":"crossref","first-page":"162","DOI":"10.1186\/1471-2105-15-162","article-title":"A multivariate approach to the integration of multi-omics datasets","volume":"15","author":"Meng","year":"2014","journal-title":"BMC Bioinformatics"},{"key":"2023112916385421400_ref12","doi-asserted-by":"crossref","first-page":"241","DOI":"10.1016\/j.jmva.2018.03.008","article-title":"Angle-based joint and individual variation explained","volume":"166","author":"Feng","year":"2018","journal-title":"Journal of Multivariate Analysis"},{"issue":"12","key":"2023112916385421400_ref13","doi-asserted-by":"crossref","first-page":"1053","DOI":"10.1038\/s41592-018-0229-2","article-title":"Deep generative modeling for single-cell transcriptomics","volume":"15","author":"Lopez","year":"2018","journal-title":"Nat Methods"},{"issue":"10","key":"2023112916385421400_ref14","doi-asserted-by":"crossref","first-page":"613","DOI":"10.1038\/nrg.2017.47","article-title":"Synthetic lethality and cancer","volume":"18","author":"O\u2019Neil","year":"2017","journal-title":"Nat Rev Genet"},{"key":"2023112916385421400_ref15","doi-asserted-by":"crossref","first-page":"e65381","DOI":"10.7554\/eLife.65381","article-title":"Evidence for additive and synergistic action of mammalian enhancers during cell fate determination","volume":"10","author":"Choi","year":"2021","journal-title":"Elife"},{"key":"2023112916385421400_ref16","article-title":"Auto-encoding variational Bayes","volume-title":"2nd International Conference on Learning Representations","author":"Kingma","year":"2014"},{"issue":"3","key":"2023112916385421400_ref17","doi-asserted-by":"crossref","first-page":"272","DOI":"10.1038\/s41592-020-01050-x","article-title":"Joint probabilistic modeling of single-cell multi-omic data with totalVI","volume":"18","author":"Gayoso","year":"2021","journal-title":"Nat Methods"},{"issue":"1","key":"2023112916385421400_ref18","doi-asserted-by":"crossref","first-page":"7419","DOI":"10.1038\/s41467-022-35094-8","article-title":"A unified computational framework for single-cell data integration with optimal transport","volume":"13","author":"Cao","year":"2022","journal-title":"Nat Commun"},{"key":"2023112916385421400_ref19","article-title":"Multimodal generative models for scalable weakly-supervised learning","author":"Wu","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2023112916385421400_ref20","article-title":"Variational mixture-of-experts autoencoders for multi-modal deep generative models","volume-title":"Advances in Neural Information Processing Systems","author":"Shi"},{"key":"2023112916385421400_ref21","volume-title":"Multimodal variational autoencoders for semi-supervised learning: In defense of product-of-experts","author":"Kutuzova"},{"key":"2023112916385421400_ref22","doi-asserted-by":"crossref","article-title":"Multicpa: multimodal compositional perturbation autoencoder","author":"Inecik","DOI":"10.1101\/2022.07.08.499049"},{"key":"2023112916385421400_ref23","volume-title":"scMM: mixture-of-experts multimodal deep generative model for single-cell multiomics data analysis","author":"Minoura"},{"issue":"12","key":"2023112916385421400_ref24","doi-asserted-by":"crossref","first-page":"1452","DOI":"10.1038\/s41587-019-0290-0","article-title":"High-throughput sequencing of the transcriptome and chromatin accessibility in the same cell","volume":"37","author":"Chen","year":"2019","journal-title":"Nat Biotechnol"},{"issue":"10","key":"2023112916385421400_ref25","doi-asserted-by":"crossref","first-page":"1243","DOI":"10.1038\/s41592-022-01601-4","article-title":"ISSAAC-seq enables sensitive and flexible multimodal profiling of chromatin accessibility and gene expression in single cells","volume":"19","author":"Xu","year":"2022","journal-title":"Nat Methods"},{"issue":"5","key":"2023112916385421400_ref26","doi-asserted-by":"crossref","first-page":"904","DOI":"10.1038\/s41591-021-01329-2","article-title":"Single-cell multi-omics analysis of the immune response in COVID-19","volume":"27","author":"Stephenson","year":"2021","journal-title":"Nat Med"},{"key":"2023112916385421400_ref27","doi-asserted-by":"crossref","DOI":"10.3389\/fmolb.2022.962644","article-title":"The performance of deep generative models for learning joint embeddings of single-cell multi-omics data","volume":"9","author":"Brombacher","year":"2022","journal-title":"Front Mol Biosci"},{"key":"2023112916385421400_ref28","doi-asserted-by":"crossref","first-page":"01","DOI":"10.1093\/bib\/bbab579","article-title":"Accurate and fast cell marker gene identification with COSG","volume":"23","author":"Dai","year":"2022","journal-title":"Brief Bioinform"},{"key":"2023112916385421400_ref29","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-031-29119-7_8","article-title":"Percolate: an exponential family jive model to design dna-based predictors of drug response","volume-title":"Research in Computational Molecular Biology","author":"Mourragui"},{"key":"2023112916385421400_ref30","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0292126","article-title":"Benchmarking variational autoencoders on cancer transcriptomics data","volume-title":"PLoS ONE","author":"Eltager"},{"key":"2023112916385421400_ref31","article-title":"Semi-supervised learning with deep generative models","volume-title":"Advances in Neural Information Processing Systems","author":"Kingma"},{"key":"2023112916385421400_ref32","doi-asserted-by":"crossref","DOI":"10.1109\/ICASSP49357.2023.10094856","article-title":"Semi-supervised classification using a supervised autoencoder for biomedical applications","volume-title":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","author":"Gille"},{"issue":"6","key":"2023112916385421400_ref33","doi-asserted-by":"crossref","first-page":"2350","DOI":"10.1214\/aos\/1032181158","article-title":"Heuristics of instability and stabilization in model selection","volume":"24","author":"Breiman","year":"1996","journal-title":"Ann Stat"},{"issue":"1","key":"2023112916385421400_ref34","doi-asserted-by":"crossref","first-page":"41","DOI":"10.1109\/TPAMI.2014.2343973","article-title":"Data fusion by matrix factorization","volume":"37","author":"\u017eitnik","year":"2015","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"2023112916385421400_ref35","first-page":"699","article-title":"2016 12th International Conference on Signal-Image Technology & Internet-Based Systems (SITIS)","volume-title":"A GPU-accelerated SVD algorithm, based on QR factorization and givens rotations, for DWI denoising","author":"Marcellino","year":"2016"},{"key":"2023112916385421400_ref36","article-title":"Learning structured output representation using deep conditional generative models","volume-title":"Advances in Neural Information Processing Systems","author":"Sohn","year":"2015"},{"key":"2023112916385421400_ref37","article-title":"Doubly reparameterized gradient estimators for Monte Carlo objectives","author":"Tucker","year":"2018"},{"issue":"10","key":"2023112916385421400_ref38","doi-asserted-by":"crossref","first-page":"1113","DOI":"10.1038\/ng.2764","article-title":"The cancer genome atlas pan-cancer analysis project","volume":"45","author":"Chang","year":"2013","journal-title":"Nat Genet"},{"key":"2023112916385421400_ref39","volume-title":"Pan-cancer atlas dataset: gene expression rnaseq - batch effects normalized mrna data","author":"The Cancer Genome Atlas"},{"key":"2023112916385421400_ref40","volume-title":"Pan-cancer atlas dataset: copy number (gene-level) - gene-level copy number (gistic2)","author":"The Cancer Genome Atlas"},{"issue":"4","key":"2023112916385421400_ref41","doi-asserted-by":"crossref","first-page":"R41","DOI":"10.1186\/gb-2011-12-4-r41","article-title":"Gistic2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers","volume":"12","author":"Mermel","year":"2011","journal-title":"Genome Biol"},{"key":"2023112916385421400_ref42","article-title":"Pan-cancer atlas dataset: Dna methylation - dna methylation (methylation450k)","author":"The Cancer Genome Atlas"},{"key":"2023112916385421400_ref43","article-title":"Pan-cancer atlas dataset: phenotype - curated clinical data","author":"The Cancer Genome Atlas"},{"issue":"2","key":"2023112916385421400_ref44","doi-asserted-by":"crossref","first-page":"163","DOI":"10.1038\/s41587-021-01206-w","article-title":"A python library for probabilistic analysis of single-cell omics data","volume":"40","author":"Gayoso","year":"2022","journal-title":"Nat Biotechnol"},{"issue":"2","key":"2023112916385421400_ref45","doi-asserted-by":"crossref","first-page":"400","DOI":"10.1016\/j.cell.2018.02.052","article-title":"An integrated TCGA pan-cancer clinical data resource to drive high-quality survival outcome analytics","volume":"173","author":"Liu","year":"2018","journal-title":"Cell"},{"key":"2023112916385421400_ref46","first-page":"199","volume-title":"Information Theory and an Extension of the Maximum Likelihood Principle","author":"Akaike","year":"1998"},{"issue":"1","key":"2023112916385421400_ref47","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1111\/j.2517-6161.1977.tb01603.x","article-title":"An asymptotic equivalence of choice of model by cross-validation and Akaike\u2019s criterion","volume":"39","author":"Stone","year":"1977","journal-title":"J R Stat Soc B Methodol"}],"container-title":["Briefings in Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bib\/article-pdf\/25\/1\/bbad416\/53896396\/bbad416.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bib\/article-pdf\/25\/1\/bbad416\/53896396\/bbad416.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,11,1]],"date-time":"2024-11-01T00:14:16Z","timestamp":1730420056000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bib\/article\/doi\/10.1093\/bib\/bbad416\/7450271"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,11,22]]},"references-count":47,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2023,11,22]]}},"URL":"https:\/\/doi.org\/10.1093\/bib\/bbad416","relation":{"has-preprint":[{"id-type":"doi","id":"10.1101\/2023.04.10.535672","asserted-by":"object"}]},"ISSN":["1467-5463","1477-4054"],"issn-type":[{"value":"1467-5463","type":"print"},{"value":"1477-4054","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2024,1,1]]},"published":{"date-parts":[[2023,11,22]]},"article-number":"bbad416"}}