Scalable Amortized GPLVMs for Single Cell Transcriptomics Data

Sarah Zhao; Aditya Ravuri; Vidhi Lalchand; Neil D. Lawrence

Back to publications

Scalable Amortized GPLVMs for Single Cell Transcriptomics Data

Sarah Zhao, Aditya Ravuri, Vidhi Lalchand, Neil D. Lawrence

, 2024.

Abstract

Dimensionality reduction is crucial for analyzing large-scale single-cell RNA-seq data. Gaussian Process Latent Variable Models (GPLVMs) offer an interpretable dimensionality reduction method, but current scalable models lack effectiveness in clustering cell types. We introduce an improved model, the amortized stochastic variational Bayesian GPLVM (BGPLVM), tailored for single-cell RNA-seq with specialized encoder, kernel, and likelihood designs. This model matches the performance of the leading single-cell variational inference (scVI) approach on synthetic and real-world COVID datasets and effectively incorporates cell-cycle and batch information to reveal more interpretable latent structures as we demonstrate on an innate immunity dataset.

Links

Cite this Paper

BibTeX


@Misc{publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data,
  title = 	 {Scalable Amortized GPLVMs for Single Cell Transcriptomics Data},
  author = 	 {Zhao, Sarah and Ravuri, Aditya and Lalchand, Vidhi and Lawrence, Neil D.},
  year = 	 {2024},
  url = 	 {/publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data.html},
  abstract = 	 {Dimensionality reduction is crucial for analyzing large-scale single-cell RNA-seq data. Gaussian Process Latent Variable Models (GPLVMs) offer an interpretable dimensionality reduction method, but current scalable models lack effectiveness in clustering cell types. We introduce an improved model, the amortized stochastic variational Bayesian GPLVM (BGPLVM), tailored for single-cell RNA-seq with specialized encoder, kernel, and likelihood designs. This model matches the performance of the leading single-cell variational inference (scVI) approach on synthetic and real-world COVID datasets and effectively incorporates cell-cycle and batch information to reveal more interpretable latent structures as we demonstrate on an innate immunity dataset.
}
}

Endnote

%0 Generic
%T Scalable Amortized GPLVMs for Single Cell Transcriptomics Data
%A Sarah Zhao
%A Aditya Ravuri
%A Vidhi Lalchand
%A Neil D. Lawrence
%D 2024	
%F publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data
%U /publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data.html
%X Dimensionality reduction is crucial for analyzing large-scale single-cell RNA-seq data. Gaussian Process Latent Variable Models (GPLVMs) offer an interpretable dimensionality reduction method, but current scalable models lack effectiveness in clustering cell types. We introduce an improved model, the amortized stochastic variational Bayesian GPLVM (BGPLVM), tailored for single-cell RNA-seq with specialized encoder, kernel, and likelihood designs. This model matches the performance of the leading single-cell variational inference (scVI) approach on synthetic and real-world COVID datasets and effectively incorporates cell-cycle and batch information to reveal more interpretable latent structures as we demonstrate on an innate immunity dataset.

RIS


TY  - GEN
TI  - Scalable Amortized GPLVMs for Single Cell Transcriptomics Data
AU  - Sarah Zhao
AU  - Aditya Ravuri
AU  - Vidhi Lalchand
AU  - Neil D. Lawrence
DA  - 2024/05/06	
ID  - publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data
UR  - /publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data.html
AB  - Dimensionality reduction is crucial for analyzing large-scale single-cell RNA-seq data. Gaussian Process Latent Variable Models (GPLVMs) offer an interpretable dimensionality reduction method, but current scalable models lack effectiveness in clustering cell types. We introduce an improved model, the amortized stochastic variational Bayesian GPLVM (BGPLVM), tailored for single-cell RNA-seq with specialized encoder, kernel, and likelihood designs. This model matches the performance of the leading single-cell variational inference (scVI) approach on synthetic and real-world COVID datasets and effectively incorporates cell-cycle and batch information to reveal more interpretable latent structures as we demonstrate on an innate immunity dataset.

ER  -

APA


Zhao, S., Ravuri, A., Lalchand, V. & Lawrence, N.D.. (2024). Scalable Amortized GPLVMs for Single Cell Transcriptomics Data.  Available from /publications/scalable-amortized-gplvms-for-single-cell-transcriptomics-data.html.