High-resolution digital dissociation of brain tumors with deep multimodal autoencoder DOI Creative Commons

Jiao Sun,

Yue Pan, Tong Lin

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 3, 2025

Abstract Single-cell technologies enable high-resolution, multi-dimensional analysis of molecular profiles in cancer biology but face challenges related to low coverage and cell annotation. The inherent hetero-geneity complexity brain tumors may hinder large-scale single multi-omic profiling. An efficient alternative is digital dissociation, which involves quantifying abundance purifying bulk samples at high resolution. However, most existing tools for resolving transcriptomes using scRNA-seq as a reference are not easily transferred other omics (e.g., chromatin accessibility, DNA methylation, protein) due ambiguous markers. Here, we introduce MODE, novel multimodal autoencoder neural network designed jointly recover personalized estimate cellular compositions. MODE the first algorithm trained on pseudo-bulk multi-omics derived from an external individualized non-RNA panel constructed target tumors. accuracy was evaluated through extensive simulation study, generated realistic data distinct tissue types. outperformed deconvolution pipelines with superior generalizability. Additionally, high-resolution purified by showed strong fidelity enhanced power detect differentially expressed genes. We applied methylome-transcriptome two independent tumor cohorts, revealing modality-specific landscapes pediatric medul-loblastoma (MB) adult glioblastoma (GBM). In MB tumors, accurately predicted composition embryonal lineage cells their marker genes expression. GBM, deconvoluted revealed increased myeloid associated poorer event-free survival. Overall, dissociation unravels origins, evolution, prognosis offering powerful tool state resolution without sequencing. pipeline freely available https://github.com/jsuncompubio/MODE .

Language: Английский

Onco-Breastomics: An Eco-Evo-Devo Holistic Approach DOI Open Access

Anca-Narcisa Neagu,

Danielle Whitham, Pathea Bruno

et al.

International Journal of Molecular Sciences, Journal Year: 2024, Volume and Issue: 25(3), P. 1628 - 1628

Published: Jan. 28, 2024

Known as a diverse collection of neoplastic diseases, breast cancer (BC) can be hyperbolically characterized dynamic pseudo-organ, living organism able to build complex, open, hierarchically organized, self-sustainable, and self-renewable tumor system, population, species, local community, biocenosis, or an evolving dynamical ecosystem (i.e., immune metabolic ecosystem) that emphasizes both developmental continuity spatio-temporal change. Moreover, cell also known oncobiota, has been described non-sexually reproducing well migratory invasive species expresses intelligent behavior, endangered parasite fights survive, optimize its features inside the host’s ecosystem, is exploit disrupt host circadian cycle for improving own proliferation spreading. BC tumorigenesis compared with early embryo placenta development may suggest new strategies research therapy. Furthermore, environmental disease ecological disorder. Many mechanisms progression have explained by principles ecology, biology, evolutionary paradigms. authors discussed ecological, developmental, more successful anti-cancer therapies, understanding bases exploitable vulnerabilities. Herein, we used integrated framework three theories: Bronfenbrenner’s theory human development, Vannote’s River Continuum Concept (RCC), Ecological Evolutionary Developmental Biology (Eco-Evo-Devo) theory, explain understand several eco-evo-devo-based govern progression. Multi-omics fields, taken together onco-breastomics, offer better opportunities integrate, analyze, interpret large amounts complex heterogeneous data, such various big-omics data obtained multiple investigative modalities, drive treatment. These integrative eco-evo-devo theories help clinicians diagnose treat BC, example, using non-invasive biomarkers in liquid-biopsies emerged from omics-based accurately reflect biomolecular landscape primary order avoid mutilating preventive surgery, like bilateral mastectomy. From perspective preventive, personalized, participatory medicine, these hypotheses patients think about this process governed natural rules, possible causes disease, gain control on their health.

Language: Английский

Citations

4
Single-Cell Transcription Mapping of Murine and Human Mammary Organoids Responses to Female Hormones DOI Creative Commons

Jenelys Ruiz Ortiz,

Steven M. Lewis,

Michael F. Ciccone

et al.

Journal of Mammary Gland Biology and Neoplasia, Journal Year: 2024, Volume and Issue: 29(1)

Published: Jan. 30, 2024

During female adolescence and pregnancy, rising levels of hormones result in a cyclic source signals that control the development mammary tissue. While such alterations are well understood from whole-gland perspective, bring to organoid cultures derived glands have yet be fully mapped. This is special importance given organoids considered suitable systems understand cross species breast development. Here we utilized single-cell transcriptional profiling delineate responses murine human normal across evolutionary distinct species. Collectively, our study represents molecular atlas epithelial dynamics response estrogen pregnancy hormones.

Language: Английский

Citations

4
Dissection of triple-negative breast cancer microenvironment and identification of potential therapeutic drugs using single-cell RNA sequencing analysis DOI Creative Commons
Weilun Cheng, Wanqi Mi, Shiyuan Wang

et al.

Journal of Pharmaceutical Analysis, Journal Year: 2024, Volume and Issue: 14(8), P. 100975 - 100975

Published: April 2, 2024

Breast cancer remains a leading cause of mortality in women worldwide. Triple-negative breast (TNBC) is particularly aggressive subtype characterized by rapid progression, poor prognosis, and lack clear therapeutic targets. In the clinic, delineation tumor heterogeneity development effective drugs continue to pose considerable challenges. Within scope our study, high inherent was uncovered based on landscape constructed from both healthy tissue samples. Notably, TNBC exhibited significant specificity regarding cell proliferation, differentiation, disease progression. Significant associations between grade, oncogenes were established via pseudotime trajectory analysis. Consequently, we further performed comprehensive characterization microenvironment. A crucial epithelial subcluster, E8, identified as highly malignant strongly associated with proliferation TNBC. Additionally, epithelial-mesenchymal transition-associated fibroblast M2 macrophage subclusters exerted an influence E8 through cellular interactions, contributing growth. Characteristic genes these three cluster cells could therefore serve potential targets for The collective findings provided valuable insights that assisted screening series drugs, such pelitinib. We confirmed anti-cancer effect pelitinib orthotopic 4T1 tumor-bearing mouse model. Overall, study sheds light unique characteristics at single-cell resolution types may potent tools drugs.

Language: Английский

Citations

4
Volumetric analysis of the terminal ductal lobular unit architecture and cell phenotypes in the human breast DOI Creative Commons
Oona Paavolainen,

Markus Peurla,

Leena M. Koskinen

et al.

Cell Reports, Journal Year: 2024, Volume and Issue: 43(10), P. 114837 - 114837

Published: Oct. 1, 2024

Language: Английский

Citations

4
High-resolution digital dissociation of brain tumors with deep multimodal autoencoder DOI Creative Commons

Jiao Sun,

Yue Pan, Tong Lin

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 3, 2025

Abstract Single-cell technologies enable high-resolution, multi-dimensional analysis of molecular profiles in cancer biology but face challenges related to low coverage and cell annotation. The inherent hetero-geneity complexity brain tumors may hinder large-scale single multi-omic profiling. An efficient alternative is digital dissociation, which involves quantifying abundance purifying bulk samples at high resolution. However, most existing tools for resolving transcriptomes using scRNA-seq as a reference are not easily transferred other omics (e.g., chromatin accessibility, DNA methylation, protein) due ambiguous markers. Here, we introduce MODE, novel multimodal autoencoder neural network designed jointly recover personalized estimate cellular compositions. MODE the first algorithm trained on pseudo-bulk multi-omics derived from an external individualized non-RNA panel constructed target tumors. accuracy was evaluated through extensive simulation study, generated realistic data distinct tissue types. outperformed deconvolution pipelines with superior generalizability. Additionally, high-resolution purified by showed strong fidelity enhanced power detect differentially expressed genes. We applied methylome-transcriptome two independent tumor cohorts, revealing modality-specific landscapes pediatric medul-loblastoma (MB) adult glioblastoma (GBM). In MB tumors, accurately predicted composition embryonal lineage cells their marker genes expression. GBM, deconvoluted revealed increased myeloid associated poorer event-free survival. Overall, dissociation unravels origins, evolution, prognosis offering powerful tool state resolution without sequencing. pipeline freely available https://github.com/jsuncompubio/MODE .

Language: Английский

Citations

0