Thomas Thorne
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Thomas Thorne has a total of 131 co-authors. Here are the top co-authors by number of shared publications:
- Michael P. H. Stumpf
School of Mathematics and Statistics, University of Melbourne, Melbourne, Australia
11 shared publications - Paul D. W. Kirk
MRC Biostatistics Unit
4 shared publications - Carsten Wiuf
Department of Mathematical Sciences, University of Copenhagen
3 shared publications Michael P. H. Stumpf
Centre of Integrative Systems Biology and Bioinformatics, Division of Molecular Biosciences, Imperial College London, London SW7 2AZ, UK
3 shared publications- Alistair J. P. Brown
MRC Centre for Medical Mycology, University of Exeter, Geoffrey Pope Building, Stocker Road, Exeter EX4 4QD, UK
2 shared publications - Chris Barnes
University College London
2 shared publications
Latest Publications
- Open Access
- 27 September 2026
- Published by Springer Nature in Statistics and Computing
- Vol. 36 (5) , 225
Rare event ABC-SMC$^{2}$Approximate Bayesian computation (ABC) is a well-established family of Monte Carlo methods for performing approximate Bayesian inference in the case where an “implicit” model is used for the data: when the data model can be simulated, but the likelihood cannot easily be pointwise evaluated. A fundamental property of standard ABC approaches is that the number of Monte Carlo points required to achieve a given accuracy scales exponentially with the dimension of the data. Prangle et al. (2018) proposes a Markov chain Monte Carlo (MCMC) method that uses a rare event sequential Monte Carlo (SMC) approach to estimating the ABC likelihood that avoids this exponential scaling, and thus allows ABC to be used on higher dimensional data. This paper builds on the work of Prangle et al. (2018) by using the rare event SMC approach within an SMC algorithm, instead of within an MCMC algorithm. The new method has a similar structure to SMC $$^{2}$$ (Chopin et al. 2013), and requires less tuning than the MCMC approach. We demonstrate the new approach, compared to existing ABC-SMC methods, on a toy example and on a duplication-divergence random graph model used for modelling protein interaction networks.- 27 September 2026
- Published by Springer Nature in Statistics and Computing
- Vol. 36 (5) , 225
Open AccessRare event ABC-SMC$^{2}$Approximate Bayesian computation (ABC) is a well-established family of Monte Carlo methods for performing approximate Bayesian inference in the case where an “implicit” model is used for the data: when the data model can be simulated, but the likelihood cannot easily be pointwise evaluated. A fundamental property of standard ABC approaches is that the number of Monte Carlo points required to achieve a given accuracy scales exponentially with the dimension of the data. Prangle et al. (2018) proposes a Markov chain Monte Carlo (MCMC) method that uses a rare event sequential Monte Carlo (SMC) approach to estimating the ABC likelihood that avoids this exponential scaling, and thus allows ABC to be used on higher dimensional data. This paper builds on the work of Prangle et al. (2018) by using the rare event SMC approach within an SMC algorithm, instead of within an MCMC algorithm. The new method has a similar structure to SMC $$^{2}$$ (Chopin et al. 2013), and requires less tuning than the MCMC approach. We demonstrate the new approach, compared to existing ABC-SMC methods, on a toy example and on a duplication-divergence random graph model used for modelling protein interaction networks. - Open Access
- 28 February 2026
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 42 (3)
Topological model selection: a case-study in tumour-induced angiogenesis
Comparing mathematical models offers a means to evaluate competing scientific theories. However, exact methods of model calibration are not applicable to many probabilistic models which simulate high-dimensional spatio-temporal data. Approximate Bayesian Computation is a widely-used method for parameter inference and model selection in such scenarios, and it may be combined with Topological Data Analysis to study models which simulate data with fine spatial structure. We develop a flexible pipeline for parameter inference and model selection in spatio-temporal models. Our pipeline identifies topological summary statistics which quantify spatio-temporal data and uses them to approximate parameter and model posterior distributions. We validate our pipeline on models of tumour-induced angiogenesis, inferring four parameters in three established models and identifying the correct model in synthetic test-cases. Simulation code for all models, data analyses, parameter inference and model selection is available online at https://github.com/rmcdomaths/tms/ and archived at https://doi.org/10.5281/zenodo.17392787. Supplementary Information will be available online.
- 28 February 2026
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 42 (3)
Open AccessTopological model selection: a case-study in tumour-induced angiogenesis
Comparing mathematical models offers a means to evaluate competing scientific theories. However, exact methods of model calibration are not applicable to many probabilistic models which simulate high-dimensional spatio-temporal data. Approximate Bayesian Computation is a widely-used method for parameter inference and model selection in such scenarios, and it may be combined with Topological Data Analysis to study models which simulate data with fine spatial structure. We develop a flexible pipeline for parameter inference and model selection in spatio-temporal models. Our pipeline identifies topological summary statistics which quantify spatio-temporal data and uses them to approximate parameter and model posterior distributions. We validate our pipeline on models of tumour-induced angiogenesis, inferring four parameters in three established models and identifying the correct model in synthetic test-cases. Simulation code for all models, data analyses, parameter inference and model selection is available online at https://github.com/rmcdomaths/tms/ and archived at https://doi.org/10.5281/zenodo.17392787. Supplementary Information will be available online.
- Open Access
Joint graphical model estimation using Stein-type shrinkage for fast large scale network inference in scRNAseq data
Graphical modeling is a widely used tool for analyzing conditional dependencies between variables and traditional methods may struggle to capture shared and distinct structures in multi-group or multi-condition settings. Joint graphical modeling (JGM) extends this framework by simultaneously estimating network structures across multiple related datasets, allowing for a deeper understanding of commonalities and differences. This capability is particularly valuable in fields such as genomics and neuroscience, where identifying variations in network topology can provide critical biological insights. Existing JGM methodologies largely fall into two categories: regularization-based approaches, which introduce additional penalties to enforce structured sparsity, and Bayesian frameworks, which incorporate prior knowledge to improve network inference. In this study, we explore an alternative method based on two-target linear covariance matrix shrinkage. Formula for optimal shrinkage intensities is proposed which leads to the development of JointStein framework. Performance of JointStein framework is proposed through simulation benchmarking which demonstrates its effectiveness for large-scale single-cell RNA sequencing (scRNA-seq) data analysis. Finally, we apply our approach to glioblastoma scRNA-seq data, uncovering dynamic shifts in T cell network structures across disease progression stages. The result highlights potential of JointStein framework in extracting biologically meaningful insights from high-dimensional data.
Open AccessJoint graphical model estimation using Stein-type shrinkage for fast large scale network inference in scRNAseq data
Graphical modeling is a widely used tool for analyzing conditional dependencies between variables and traditional methods may struggle to capture shared and distinct structures in multi-group or multi-condition settings. Joint graphical modeling (JGM) extends this framework by simultaneously estimating network structures across multiple related datasets, allowing for a deeper understanding of commonalities and differences. This capability is particularly valuable in fields such as genomics and neuroscience, where identifying variations in network topology can provide critical biological insights. Existing JGM methodologies largely fall into two categories: regularization-based approaches, which introduce additional penalties to enforce structured sparsity, and Bayesian frameworks, which incorporate prior knowledge to improve network inference. In this study, we explore an alternative method based on two-target linear covariance matrix shrinkage. Formula for optimal shrinkage intensities is proposed which leads to the development of JointStein framework. Performance of JointStein framework is proposed through simulation benchmarking which demonstrates its effectiveness for large-scale single-cell RNA sequencing (scRNA-seq) data analysis. Finally, we apply our approach to glioblastoma scRNA-seq data, uncovering dynamic shifts in T cell network structures across disease progression stages. The result highlights potential of JointStein framework in extracting biologically meaningful insights from high-dimensional data.
- Open Access
- 26 October 2024
- Published by Springer Nature in BMC Bioinformatics
- Vol. 25 (1) , 1-16
Shrinkage estimation of gene interaction networks in single-cell RNA sequencing data
Background: Gene interaction networks are graphs in which nodes represent genes and edges represent functional interactions between them. These interactions can be at multiple levels, for instance, gene regulation, protein-protein interaction, or metabolic pathways. To analyse gene interaction networks at a large scale, gene co-expression network analysis is often applied on high-throughput gene expression data such as RNA sequencing data. With the advance in sequencing technology, expression of genes can be measured in individual cells. Single-cell RNA sequencing (scRNAseq) provides insights of cellular development, differentiation and characteristics at the transcriptomic level. High sparsity and high-dimensional data structures pose challenges in scRNAseq data analysis. Results: In this study, a sparse inverse covariance matrix estimation framework for scRNAseq data is developed to capture direct functional interactions between genes. Comparative analyses highlight high performance and fast computation of Stein-type shrinkage in high-dimensional data using simulated scRNAseq data. Data transformation approaches also show improvement in performance of shrinkage methods in non-Gaussian distributed data. Zero-inflated modelling of scRNAseq data based on a negative binomial distribution enhances shrinkage performance in zero-inflated data without interference on non zero-inflated count data. Conclusion: The proposed framework broadens application of graphical model in scRNAseq analysis with flexibility in sparsity of count data resulting from dropout events, high performance, and fast computational time. Implementation of the framework is in a reproducible Snakemake workflow https://github.com/calathea24/ZINBGraphicalModel and R package ZINBStein https://github.com/calathea24/ZINBStein.
- 26 October 2024
- Published by Springer Nature in BMC Bioinformatics
- Vol. 25 (1) , 1-16
Open AccessShrinkage estimation of gene interaction networks in single-cell RNA sequencing data
Background: Gene interaction networks are graphs in which nodes represent genes and edges represent functional interactions between them. These interactions can be at multiple levels, for instance, gene regulation, protein-protein interaction, or metabolic pathways. To analyse gene interaction networks at a large scale, gene co-expression network analysis is often applied on high-throughput gene expression data such as RNA sequencing data. With the advance in sequencing technology, expression of genes can be measured in individual cells. Single-cell RNA sequencing (scRNAseq) provides insights of cellular development, differentiation and characteristics at the transcriptomic level. High sparsity and high-dimensional data structures pose challenges in scRNAseq data analysis. Results: In this study, a sparse inverse covariance matrix estimation framework for scRNAseq data is developed to capture direct functional interactions between genes. Comparative analyses highlight high performance and fast computation of Stein-type shrinkage in high-dimensional data using simulated scRNAseq data. Data transformation approaches also show improvement in performance of shrinkage methods in non-Gaussian distributed data. Zero-inflated modelling of scRNAseq data based on a negative binomial distribution enhances shrinkage performance in zero-inflated data without interference on non zero-inflated count data. Conclusion: The proposed framework broadens application of graphical model in scRNAseq analysis with flexibility in sparsity of count data resulting from dropout events, high performance, and fast computational time. Implementation of the framework is in a reproducible Snakemake workflow https://github.com/calathea24/ZINBGraphicalModel and R package ZINBStein https://github.com/calathea24/ZINBStein.
- Open Access
- 8 October 2024
- Published by Springer Nature in Journal of Mathematical Biology
- Vol. 89 (5) , 1-40
Calibration of stochastic, agent-based neuron growth models with approximate Bayesian computation
Understanding how genetically encoded rules drive and guide complex neuronal growth processes is essential to comprehending the brain’s architecture, and agent-based models (ABMs) offer a powerful simulation approach to further develop this understanding. However, accurately calibrating these models remains a challenge. Here, we present a novel application of Approximate Bayesian Computation (ABC) to address this issue. ABMs are based on parametrized stochastic rules that describe the time evolution of small components–the so-called agents–discretizing the system, leading to stochastic simulations that require appropriate treatment. Mathematically, the calibration defines a stochastic inverse problem. We propose to address it in a Bayesian setting using ABC. We facilitate the repeated comparison between data and simulations by quantifying the morphological information of single neurons with so-called morphometrics and resort to statistical distances to measure discrepancies between populations thereof. We conduct experiments on synthetic as well as experimental data. We find that ABC utilizing Sequential Monte Carlo sampling and the Wasserstein distance finds accurate posterior parameter distributions for representative ABMs. We further demonstrate that these ABMs capture specific features of pyramidal cells of the hippocampus (CA1). Overall, this work establishes a robust framework for calibrating agent-based neuronal growth models and opens the door for future investigations using Bayesian techniques for model building, verification, and adequacy assessment.
- 8 October 2024
- Published by Springer Nature in Journal of Mathematical Biology
- Vol. 89 (5) , 1-40
Open AccessCalibration of stochastic, agent-based neuron growth models with approximate Bayesian computation
Understanding how genetically encoded rules drive and guide complex neuronal growth processes is essential to comprehending the brain’s architecture, and agent-based models (ABMs) offer a powerful simulation approach to further develop this understanding. However, accurately calibrating these models remains a challenge. Here, we present a novel application of Approximate Bayesian Computation (ABC) to address this issue. ABMs are based on parametrized stochastic rules that describe the time evolution of small components–the so-called agents–discretizing the system, leading to stochastic simulations that require appropriate treatment. Mathematically, the calibration defines a stochastic inverse problem. We propose to address it in a Bayesian setting using ABC. We facilitate the repeated comparison between data and simulations by quantifying the morphological information of single neurons with so-called morphometrics and resort to statistical distances to measure discrepancies between populations thereof. We conduct experiments on synthetic as well as experimental data. We find that ABC utilizing Sequential Monte Carlo sampling and the Wasserstein distance finds accurate posterior parameter distributions for representative ABMs. We further demonstrate that these ABMs capture specific features of pyramidal cells of the hippocampus (CA1). Overall, this work establishes a robust framework for calibrating agent-based neuronal growth models and opens the door for future investigations using Bayesian techniques for model building, verification, and adequacy assessment.
- Open Access
Shrinkage estimation of gene interaction networks in single-cell RNA sequencing data
Gene interaction networks are graphs in which nodes represent genes and edges represent functional interactions between them. These interactions can be at multiple levels, for instance, gene regulation, protein-protein interaction, or metabolic pathways. To analyse gene interaction networks at a large scale, gene co-expression network analysis is often applied on high-throughput gene expression data such as RNA sequencing data. With the advance in sequencing technology, expression of genes can be measured in individual cells. Single-cell RNA sequencing (scRNAseq) provides insights of cellular development, differentiation and characteristics at transcriptomic level. High sparsity and high-dimensional data structure pose challenges in scRNAseq data analysis. In this study, a sparse inverse covariance matrix estimation framework for scRNAseq data is developed to capture direct functional interactions between genes. Comparative analyses highlight high performance and fast computation of Stein-type shrinkage in high-dimensional data using simulated scRNAseq data. Data transformation approaches also show improvement in performance of shrinkage methods in non-Gaussian distributed data. Zero-inflated modelling of scRNAseq data based on a negative binomial distribution enhances shrinkage performance in zero-inflated data without interference on non zeroinflated count data. The optimal zero-inflated Stein-type shrinkage framework is applied on experimental scRNAseq data which demonstrates its potential to construct sparser gene interaction networks with higher precision. Availability and implementation https://github.com/calathea24/ZINBGraphicalModel
Open AccessShrinkage estimation of gene interaction networks in single-cell RNA sequencing data
Gene interaction networks are graphs in which nodes represent genes and edges represent functional interactions between them. These interactions can be at multiple levels, for instance, gene regulation, protein-protein interaction, or metabolic pathways. To analyse gene interaction networks at a large scale, gene co-expression network analysis is often applied on high-throughput gene expression data such as RNA sequencing data. With the advance in sequencing technology, expression of genes can be measured in individual cells. Single-cell RNA sequencing (scRNAseq) provides insights of cellular development, differentiation and characteristics at transcriptomic level. High sparsity and high-dimensional data structure pose challenges in scRNAseq data analysis. In this study, a sparse inverse covariance matrix estimation framework for scRNAseq data is developed to capture direct functional interactions between genes. Comparative analyses highlight high performance and fast computation of Stein-type shrinkage in high-dimensional data using simulated scRNAseq data. Data transformation approaches also show improvement in performance of shrinkage methods in non-Gaussian distributed data. Zero-inflated modelling of scRNAseq data based on a negative binomial distribution enhances shrinkage performance in zero-inflated data without interference on non zeroinflated count data. The optimal zero-inflated Stein-type shrinkage framework is applied on experimental scRNAseq data which demonstrates its potential to construct sparser gene interaction networks with higher precision. Availability and implementation https://github.com/calathea24/ZINBGraphicalModel
- 1 October 2023
- Published by SAGE Publications in Journal of Computational Biology
- Vol. 30 (10) , 1059-1074
Identifying Subpopulations of Cells in Single-Cell Transcriptomic Data: A Bayesian Mixture Modeling Approach to Zero Inflation of Counts
In the study of single-cell RNA-seq (scRNA-Seq) data, a key component of the analysis is to identify subpopulations of cells in the data. A variety of approaches to this have been considered, and although many machine learning-based methods have been developed, these rarely give an estimate of uncertainty in the cluster assignment. To allow for this, probabilistic models have been developed, but scRNA-Seq data exhibit a phenomenon known as dropout, whereby a large proportion of the observed read counts are zero. This poses challenges in developing probabilistic models that appropriately model the data. We develop a novel Dirichlet process mixture model that employs both a mixture at the cell level to model multiple populations of cells and a zero-inflated negative binomial mixture of counts at the transcript level. By taking a Bayesian approach, we are able to model the expression of genes within clusters, and to quantify uncertainty in cluster assignments. It is shown that this approach outperforms previous approaches that applied multinomial distributions to model scRNA-Seq counts and negative binomial models that do not take into account zero inflation. Applied to a publicly available data set of scRNA-Seq counts of multiple cell types from the mouse cortex and hippocampus, we demonstrate how our approach can be used to distinguish subpopulations of cells as clusters in the data, and to identify gene sets that are indicative of membership of a subpopulation.
- 1 October 2023
- Published by SAGE Publications in Journal of Computational Biology
- Vol. 30 (10) , 1059-1074
Identifying Subpopulations of Cells in Single-Cell Transcriptomic Data: A Bayesian Mixture Modeling Approach to Zero Inflation of Counts
In the study of single-cell RNA-seq (scRNA-Seq) data, a key component of the analysis is to identify subpopulations of cells in the data. A variety of approaches to this have been considered, and although many machine learning-based methods have been developed, these rarely give an estimate of uncertainty in the cluster assignment. To allow for this, probabilistic models have been developed, but scRNA-Seq data exhibit a phenomenon known as dropout, whereby a large proportion of the observed read counts are zero. This poses challenges in developing probabilistic models that appropriately model the data. We develop a novel Dirichlet process mixture model that employs both a mixture at the cell level to model multiple populations of cells and a zero-inflated negative binomial mixture of counts at the transcript level. By taking a Bayesian approach, we are able to model the expression of genes within clusters, and to quantify uncertainty in cluster assignments. It is shown that this approach outperforms previous approaches that applied multinomial distributions to model scRNA-Seq counts and negative binomial models that do not take into account zero inflation. Applied to a publicly available data set of scRNA-Seq counts of multiple cell types from the mouse cortex and hippocampus, we demonstrate how our approach can be used to distinguish subpopulations of cells as clusters in the data, and to identify gene sets that are indicative of membership of a subpopulation.
- Open Access
- 2 August 2022
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 119 (32)
Inhibition of the angiotensin II type 2 receptor AT 2 R is a novel therapeutic strategy for glioblastoma
Glioblastoma (GBM) is an aggressive malignant primary brain tumor with limited therapeutic options. We show that the angiotensin II (AngII) type 2 receptor (AT 2 R) is a therapeutic target for GBM and that AngII, endogenously produced in GBM cells, promotes proliferation through AT 2 R. We repurposed EMA401, an AT 2 R antagonist originally developed as a peripherally restricted analgesic, for GBM and showed that it inhibits the proliferation of AT 2 R-expressing GBM spheroids and blocks their invasiveness and angiogenic capacity. The crystal structure of AT 2 R bound to EMA401 was determined and revealed the receptor to be in an active-like conformation with helix-VIII blocking G-protein or β-arrestin recruitment. The architecture and interactions of EMA401 in AT 2 R differ drastically from complexes of AT 2 R with other relevant compounds. To enhance central nervous system (CNS) penetration of EMA401, we exploited the crystal structure to design an angiopep-2–tethered EMA401 derivative, A3E. A3E exhibited enhanced CNS penetration, leading to reduced tumor volume, inhibition of proliferation, and increased levels of apoptosis in an orthotopic xenograft model of GBM.
- 2 August 2022
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 119 (32)
Open AccessInhibition of the angiotensin II type 2 receptor AT 2 R is a novel therapeutic strategy for glioblastoma
Glioblastoma (GBM) is an aggressive malignant primary brain tumor with limited therapeutic options. We show that the angiotensin II (AngII) type 2 receptor (AT 2 R) is a therapeutic target for GBM and that AngII, endogenously produced in GBM cells, promotes proliferation through AT 2 R. We repurposed EMA401, an AT 2 R antagonist originally developed as a peripherally restricted analgesic, for GBM and showed that it inhibits the proliferation of AT 2 R-expressing GBM spheroids and blocks their invasiveness and angiogenic capacity. The crystal structure of AT 2 R bound to EMA401 was determined and revealed the receptor to be in an active-like conformation with helix-VIII blocking G-protein or β-arrestin recruitment. The architecture and interactions of EMA401 in AT 2 R differ drastically from complexes of AT 2 R with other relevant compounds. To enhance central nervous system (CNS) penetration of EMA401, we exploited the crystal structure to design an angiopep-2–tethered EMA401 derivative, A3E. A3E exhibited enhanced CNS penetration, leading to reduced tumor volume, inhibition of proliferation, and increased levels of apoptosis in an orthotopic xenograft model of GBM.
- Open Access
- 22 February 2022
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 38 (9) , 2529-2535
Topological approximate Bayesian computation for parameter inference of an angiogenesis model
Motivation: Inferring the parameters of models describing biological systems is an important problem in the reverse engineering of the mechanisms underlying these systems. Much work has focused on parameter inference of stochastic and ordinary differential equation models using Approximate Bayesian Computation (ABC). While there is some recent work on inference in spatial models, this remains an open problem. Simultaneously, advances in topological data analysis (TDA), a field of computational mathematics, have enabled spatial patterns in data to be characterized.Results: Here, we focus on recent work using TDA to study different regimes of parameter space for a well-studied model of angiogenesis. We propose a method for combining TDA with ABC to infer parameters in the Anderson–Chaplain model of angiogenesis. We demonstrate that this topological approach outperforms ABC approaches that use simpler statistics based on spatial features of the data. This is a first step toward a general framework of spatial parameter inference for biological systems, for which there may be a variety of filtrations, vectorizations and summary statistics to be considered.Availability and implementation: All code used to produce our results is available as a Snakemake workflow from github.com/tt104/tabc_angio.
- 22 February 2022
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 38 (9) , 2529-2535
Open AccessTopological approximate Bayesian computation for parameter inference of an angiogenesis model
Motivation: Inferring the parameters of models describing biological systems is an important problem in the reverse engineering of the mechanisms underlying these systems. Much work has focused on parameter inference of stochastic and ordinary differential equation models using Approximate Bayesian Computation (ABC). While there is some recent work on inference in spatial models, this remains an open problem. Simultaneously, advances in topological data analysis (TDA), a field of computational mathematics, have enabled spatial patterns in data to be characterized.Results: Here, we focus on recent work using TDA to study different regimes of parameter space for a well-studied model of angiogenesis. We propose a method for combining TDA with ABC to infer parameters in the Anderson–Chaplain model of angiogenesis. We demonstrate that this topological approach outperforms ABC approaches that use simpler statistics based on spatial features of the data. This is a first step toward a general framework of spatial parameter inference for biological systems, for which there may be a variety of filtrations, vectorizations and summary statistics to be considered.Availability and implementation: All code used to produce our results is available as a Snakemake workflow from github.com/tt104/tabc_angio.
- Open Access
Identifying sub-populations of cells in single cell transcriptomic data – a Bayesian mixture modelling approach to zero-inflation of counts
In the study of single cell RNA-seq data, a key component of the analysis is to identify sub-populations of cells in the data. A variety of approaches to this have been considered, and although many machine learning based methods have been developed, these rarely give an estimate of uncertainty in the cluster assignment. To allow for this probabilistic models have been developed, but single cell RNA-seq data exhibit a phenomenon known as dropout, whereby a large proportion of the observed read counts are zero. This poses challenges in developing probabilistic models that appropriately model the data. We develop a novel Dirichlet process mixture model which employs both a mixture at the cell level to model multiple populations of cells, and a zero-inflated negative binomial mixture of counts at the transcript level. By taking a Bayesian approach we are able to model the expression of genes within clusters, and to quantify uncertainty in cluster assignments. It is shown that this approach out-performs previous approaches that applied multinomial distributions to model single cell RNA-seq counts and negative binomial models that do not take into account zero-inflation. Applied to a publicly available data set of single cell RNA-seq counts of multiple cell types from the mouse cortex and hippocampus, we demonstrate how our approach can be used to distinguish sub-populations of cells as clusters in the data, and to identify gene sets that are indicative of membership of a sub-population. The methodology is implemented as an open source Snakemake pipeline available from https://github.com/tt104/scmixture .
Open AccessIdentifying sub-populations of cells in single cell transcriptomic data – a Bayesian mixture modelling approach to zero-inflation of counts
In the study of single cell RNA-seq data, a key component of the analysis is to identify sub-populations of cells in the data. A variety of approaches to this have been considered, and although many machine learning based methods have been developed, these rarely give an estimate of uncertainty in the cluster assignment. To allow for this probabilistic models have been developed, but single cell RNA-seq data exhibit a phenomenon known as dropout, whereby a large proportion of the observed read counts are zero. This poses challenges in developing probabilistic models that appropriately model the data. We develop a novel Dirichlet process mixture model which employs both a mixture at the cell level to model multiple populations of cells, and a zero-inflated negative binomial mixture of counts at the transcript level. By taking a Bayesian approach we are able to model the expression of genes within clusters, and to quantify uncertainty in cluster assignments. It is shown that this approach out-performs previous approaches that applied multinomial distributions to model single cell RNA-seq counts and negative binomial models that do not take into account zero-inflation. Applied to a publicly available data set of single cell RNA-seq counts of multiple cell types from the mouse cortex and hippocampus, we demonstrate how our approach can be used to distinguish sub-populations of cells as clusters in the data, and to identify gene sets that are indicative of membership of a sub-population. The methodology is implemented as an open source Snakemake pipeline available from https://github.com/tt104/scmixture .