Biology subjects

Wernisch, L.

Publications and source records attributed to Wernisch, L..

7 recordsLinked to original sources

bioRxiv2018

Branch-recombinant Gaussian processes for analysis of perturbations in biological time series

MotivationA common class of behaviour encountered in the biological sciences involves branching and recombination. During branching, a statistical process bifurcates resulting in two or more potentially correlated processes that may under-go further branching; the contrary is true during recombination, where two or more statistical processes converge into one. A key objective is to identify the time of this bifurcation (branch time) from time series measurements e.g., comparing a control time series with a perturbed time series. Whilst statistical treatments for the two branch (control versus treatment) case exists, the ability to infer more complex branching structure from time series data remains open. Gaussian processes (GPs) represents an ideal framework for such analysis, allowing for nonlinear regression that includes a rigorous treatment of uncertainty. Currently, however, GP models only exist for two-branch systems. Here we highlight how arbitrarily complex branching processes can be built using the correct composition of covariance functions within a GP framework, thus outlining a general framework for the treatment of branching and recombination in the form of branch-recombinant Gaussian processes (B-RGPs). We first demonstrate the performance of B-RGPs compared to a variety of existing regression approaches, and demonstrate robustness to model misspecification. B-RGPs are then used to investigate the branching patterns of Arabidopsis thaliana gene expression following inoculation with the hemibotrophic bacteria, Pseudomonas syringae DC3000, and a disarmed mutant strain, hrpA. By grouping genes according to the number of branches, we could naturally separate out genes involved in basal immune response from those subverted by the virulent strain, and show enrichment for targets of pathogen protein effectors. Finally, we identify two early branching genes WRKY11 and WRKY17, and showed that groups of genes that branched at similar times to WRKY11/17 were enriched for W-box binding motifs, and overrepresented for genes differentially expressed in WRKY11/17 knockouts, suggesting that branch time could be used for identifying direct and indirect binding targets of key transcription factors. Software is available from: https://github.com/cap76/BranchingGPs.

bioinformatics↗

bioRxiv2017

GPseudoRank: MCMC for sampling from posterior distributions of pseudo-orderings using Gaussian processes

MotivationA number of pseudotime methods have provided point estimates of the ordering of cells for scRNA-seq data. A still limited number of methods also model the uncertainty of the pseudotime estimate. However, there is still a need for a method to sample from complicated and multi-modal distributions of orders, and to estimate changes in the amount of the uncertainty of the order during the course of a biological development, as this can support the selection of suitable cells for the clustering of genes or for network inference.\n\nResultsIn an application to a microarray data set our proposed method, GPseudoRank, identifies two modes of the distribution, each of them corresponding to point estimates of orders obtained by a different established method. In an application to scRNA-seq data we demonstrate the potential of GPseudoRank to identify phases of lower and higher pseudotime uncertainty during a biological process. GPseudoRank also correctly identifies cells precocious in their antiviral response.\n\nAvailability and implementationOur method is available on github: https://github.com/magStra/GPseudoRank.\n\nContactmagdalena.strauss@mrc-bsu.cam.ac.uk\n\nSupplementary informationSupplementary materials are available.

bioinformatics↗

bioRxiv2017

Nonparametric Bayesian inference of transcriptional branching and recombination identifies regulators of early human germ cell development

During embryonic development, cells undertake a series of fate decisions to form a complete organism comprised of various cell types, epitomising a branching process. A striking example of branching occurs in humans around the time of implantation, when primordial germ cells (PGCs), precursors of sperm and eggs, and somatic lineages are specified. Due to inaccessibility of human embryos at this stage of development, understanding the mechanisms of PGC specification remains difficult. The integrative modelling of single cell transcriptomics data from embryos and appropriate in vitro models should prove to be a useful resource for investigating this system, provided that the cells can be suitably ordered over a developmental axis. Unfortunately, most methods for inferring cell ordering were not designed with structured (time series) data in mind. Although some probabilistic approaches address these limitations by incorporating prior information about the developmental stage (capture time) of the cell, they do not allow the ordering of cells over processes with more than one terminal cell fate. To investigate the mechanisms of PGC specification, we develop a probabilistic pseudotime approach, branch-recombinant Gaussian process latent variable models (B-RGPLVMs), that use an explicit model of transcriptional branching in individual marker genes, allowing the ordering of cells over developmental trajectories with arbitrary numbers of branches. We use first demonstrate the advantage of our approach over existing pseudotime algorithms and subsequently use it to investigate early human development, as primordial germ cells (PGCs) and somatic cells diverge. We identify known master regulators of human PGCs, and predict roles for a variety of signalling pathways, transcription factors, and epigenetic modifiers. By concentrating on the earliest branched signalling events, we identified an antagonistic role for FGF receptor (FGFR) signalling pathway in the acquisition of competence for human PGC fate, and identify putative roles for PRC1 and PRC2 in PGC specification. We experimentally validate our predictions using pharmacological blocking of FGFR or its downstream effectors (MEK, PI3K and JAK), and demonstrate enhanced competency for PGC fate in vitro, whilst small molecule inhibition of the enzymatic component of PRC1/PRC2 reveals reduced capacity of cells to form PGCs in vitro. Thus, B-RGPLVMs represent a powerful and flexible data-driven approach for dissecting the temporal dynamics of cell fate decisions, providing unique insights into the mechanisms of early embryogenesis. Scripts relating to this analysis are available from: https://github.com/cap76/PGCPseudotime

developmental biology↗

bioRxiv2017

A Comparison Of Machine Learning And Bayesian Modelling For Molecular Serotyping

BackgroundStreptococcus pneumoniae is a human pathogen that is a major cause of infant mortality. Identifying the pneumococcal serotype is an important step in monitoring the impact of vaccines used to protect against disease. Genomic microarrays provide an effective method for molecular serotyping. Previously we developed an empirical Bayesian model for the classification of serotypes from a molecular serotyping array. With only few samples available, a model driven approach was the only option. In the meanwhile, several thousand samples have been made available to us, providing an opportunity to investigate serotype classification by machine learning methods, which could complement the Bayesian model.\n\nResultsWe compare the performance of the original Bayesian model with two machine learning algorithms: Gradient Boosting Machines and Random Forests. We present our results as an example of a generic strategy whereby a preliminary probabilistic model is complemented or replaced by a machine learning classifier once enough data are available. Despite the availability of thousands of serotyping arrays, a problem encountered when applying machine learning methods is the lack of training data containing mixtures of serotypes; due to the large number of possible combinations. Most of the available training data comprises samples with only a single serotype. To overcome the lack of training data we implemented an iterative analysis, creating artificial training data of serotype mixtures by combining raw data from single serotype arrays.\n\nConclusionsWith the enhanced training set the machine learning algorithms out perform the original Bayesian model. However, for serotypes currently lacking sufficient training data the best performing implementation was a combination of the results of the Bayesian Model and the Gradient Boosting Machine. As well as being an effective method for classifying biological data, machine learning can also be used as an efficient method for revealing subtle biological insights, which we illustrate with an example.

genomics↗

bioRxiv2017

Exploiting General Independence Criteria For Network Inference

Inference of networks representing dependency relationships is a key tool for understanding data derived from biological systems. It has been shown that nonlinear relationships and non-Gaussian noise aid detection of directions of functional dependencies. In this study we explore how far generalised independence criteria for statistical independence proposed in the literature are better suited to the inference of networks compared to standard independence criteria based on linear relationships and Gaussian noise. We compare such criteria within the framework of the PC algorithm, a popular network inference algorithm for directed acyclic dependency graphs. We also propose and evaluate a method to apply unconditional independence criteria to assess conditional independence and a method to simulate data with desired properties from experimental data. Our main finding is that a recently proposed criterion based on distance covariance performs well compared to other independence criteria in terms of error rates, speed of computation, and need of fine-tuning parameters when applied to experimental biological datasets.

bioinformatics↗

bioRxiv2017

Clusternomics: Integrative Context-Dependent Clustering For Heterogeneous Datasets

Integrative clustering is used to identify groups of samples by jointly analysing multiple datasets describing the same set of biological samples, such as gene expression, copy number, methylation etc. Most existing algorithms for integrative clustering assume that there is a shared consistent set of clusters across all datasets, and most of the data samples follow this structure. However in practice, the structure across heterogeneous datasets can be more varied, with clusters being joined in some datasets and separated in others.\n\nIn this paper, we present a probabilistic clustering method to identify groups across datasets that do not share the same cluster structure. The proposed algorithm, Clusternomics, identifies groups of samples that share their global behaviour across heterogeneous datasets. The algorithm models clusters on the level of individual datasets, while also extracting global structure that arises from the local cluster assignments. Clusters on both the local and the global level are modelled using a hierarchical Dirichlet mixture model to identify structure on both levels.\n\nWe evaluated the model both on simulated and on real-world datasets. The simulated data exemplifies datasets with varying degrees of common structure. In such a setting Clusternomics outperforms existing algorithms for integrative and consensus clustering. In a real-world application, we used the algorithm for cancer subtyping, identifying subtypes of cancer from heterogeneous datasets. We applied the algorithm to TCGA breast cancer dataset, integrating gene expression, miRNA expression, DNA methylation and proteomics. The algorithm extracted clinically meaningful clusters with significantly different survival probabilities. We also evaluated the algorithm on lung and kidney cancer TCGA datasets with high dimensionality, again showing clinically significant results and scalability of the algorithm.\n\nAuthor SummaryIntegrative clustering is the task of identifying groups of samples by combining information from several datasets. An example of this task is cancer subtyping, where we cluster tumour samples based on several datasets, such as gene expression, proteomics and others. Most existing algorithms assume that all such datasets share a similar cluster structure, with samples outside these clusters treated as noise. The structure can, however, be much more heterogeneous: some meaningful clusters may appear only in some datasets.\n\nIn the paper, we introduce the Clusternomics algorithm that identifies groups of samples across heterogeneous datasets. It models both cluster structure of individual datasets, and the global structure that appears as a combination of local structures. The algorithm uses probabilistic modelling to identify the groups and share information across the local and global levels. We evaluated the algorithm on both simulated and real world datasets, where the algorithm found clinically significant clusters with different survival outcomes.

genomics↗

bioRxiv2017

gene-cocite: a web application for extracting, visualising and assessing the cocitations of a list of genes

BackgroundThe outcome from the analysis of high through-put genomics experiments is commonly a list of genes. The most basic measure of association is whether the genes in the list have ever been cocited together.\n\nResultsThe web application gene-cocite accepts a list of genes and returns a list of the papers which cocite any two or more of the genes. The proportion of the genes which are cocited with at least one other gene is given, and the p-value for the probability of this proportion of cocitations occurring by chance from a random list of genes of the same length calculated. An interactive graph with links to papers is displayed, showing how the genes in the list are related to each other by publications.\n\nConclusionsgene-cocite (http://sysbio.mrc-bsu.cam.ac.uk/gene-cocite) is designed to be an easy to use first step for biological researchers investigating the background of their list of genes.

bioinformatics↗