Biology subjects

Marr, C.

Publications and source records attributed to Marr, C..

30 records · Page 2Linked to original sources

bioRxiv2020

InstantDL - An easy-to-use deep learning pipeline for image segmentation and classification

MotivationDeep learning contributes to uncovering and understanding molecular and cellular processes with highly performant image computing algorithms. Convolutional neural networks have become the state-of-the-art tool to provide accurate, consistent and fast data processing. However, published algorithms mostly solve only one specific problem and they often require expert skills and a considerable computer science and machine learning background for application. ResultsWe have thus developed a deep learning pipeline called InstantDL for four common image processing tasks: semantic segmentation, instance segmentation, pixel-wise regression and classification. InstantDL enables experts and non-experts to apply state-of-the-art deep learning algorithms to biomedical image data with minimal effort. To make the pipeline robust, we have automated and standardized workflows and extensively tested it in different scenarios. Moreover, it allows to assess the uncertainty of predictions. We have benchmarked InstantDL on seven publicly available datasets achieving competitive performance without any parameter tuning. For customization of the pipeline to specific tasks, all code is easily accessible. Availability and ImplementationInstantDL is available under the terms of MIT licence. It can be found on GitHub: https://github.com/marrlab/InstantDL Contactcarsten.marr@helmholtz-muenchen.de

bioinformatics↗

bioRxiv2020

Computational modeling reveals cell-cycle dependent kinetics of H4K20 methylation states during Xenopus embryogenesis

Histone modifications regulate chromatin architecture and thereby control gene expression. Rapid cell divisions and DNA replication however lead to a dilution of histone modifications and can thus affect chromatin mediated gene regulation So how does the cell-cycle shape the histone modification landscape, in particular during embryogenesis when a fast and precise control of cell-specific gene expression is required? We addressed this question in vivo by manipulating the cell-cycle during early Xenopus laevis embryogenesis. The global distribution of un-, mono- di- and tri-methylated histone H4K20 was measured by mass spectrometry in normal and cell-cycle arrested embryos over time. Using multi-start maximum likelihood optimization and quantitative model selection, we found that three specific methylation rate constants were required to explain the measured H4K20 methylation state kinetics. Interestingly, demethylation was found to be redundant in the cycling embryos but essential in the cell-cycle arrested embryos. Together, we present the first quantitative analysis of in vivo histone H4K20 methylation kinetics. Our computational model shows that demethylation is only essential for regulating H4K20 methylation kinetics in non-cycling cells. In rapidly dividing cells of early embryos, we predict that demethylation is dispensable, suggesting that cell-cycle mediated dilution of chromatin marks is an essential regulatory component for shaping the epigenetic landscape during early embryonic development.

bioinformatics↗

bioRxiv2020

Automatic identification of relevant genes from low-dimensional embeddings of single cell RNAseq data

Dimensionality reduction is a key step in the analysis of single-cell RNA sequencing data and produces a low-dimensional embedding for visualization and as a calculation base for downstream analysis. Nonlinear techniques are most suitable to handle the intrinsic complexity of large, heterogeneous single cell data. With no linear relation between genes and embedding however, there is no way to extract the identity of genes most relevant for any cells position in the low-dimensional embedding, and thus the underlying process. In this paper, we introduce the concepts of global and local gene relevance to compute an equivalent of principal component analysis loadings for non-linear low-dimensional embeddings. While global gene relevance identifies drivers of the overall embedding, local gene relevance singles out genes that change in small, possibly rare subsets of cells. We apply our method to single-cell RNAseq datasets from different experimental protocols and to different low dimensional embedding techniques, shows our methods versatility to identify key genes for a variety of biological processes. To ensure reproducibility and ease of use, our method is released as part of destiny 3.0, a popular R package for building diffusion maps from single-cell transcriptomic data. It is readily available through Bioconductor.

bioinformatics↗

bioRxiv2020

Aggregated spatio-temporal division patterns emerge from reoccurring divisions of neural stem cells

The regulation of quiescence and cell cycle entry is pivotal for the maintenance of stem cell populations. Regulatory mechanisms however are poorly understood. In particular it is unclear how the activity of single stem cells is coordinated within the population, or if cells divide in a purely random fashion. We addressed this issue by analyzing division events in an adult neural stem cell (NSC) population of the zebrafish telencephalon. Spatial statistics and mathematical modeling of over 80,000 NSCs in 36 brains revealed weakly aggregated, non-random division patterns in space and time. Analyzing divisions at two timepoints allowed us to infer cell cycle and S-phase lengths computationally. Interestingly, we observed rapid cell cycle re-entries in roughly 15% of newly born NSCs. In agent based simulations of NSC populations, this re-dividing activity sufficed to induce aggregated spatio-temporal division patterns that matched the ones observed experimentally. In contrast, omitting re-divisions lead to a random spatio-temporal distribution of dividing cells. Spatio-temporal aggregation of dividing stem cells can thus emerge from the cells history, regardless of possible feedback mechanisms in the population.

neuroscience↗

bioRxiv2019

A Sparse Deep Learning Approach for Automatic Segmentation of Human Vasculature in Multispectral Optoacoustic Tomography

Multispectral Optoacoustic Tomography (MSOT) resolves oxy- (HbO2) and deoxy-hemoglobin (Hb) to perform vascular imaging. MSOT suffers from gradual signal attenuation with depth due to light-tissue interactions: an effect that hinders the precise manual segmentation of vessels. Furthermore, vascular assessment requires functional tests, which last several minutes and result in recording thousands of images. Here, we introduce a deep learning approach with a sparse UNET (S-UNET) for automatic vascular segmentation in MSOT images to avoid the rigorous and time-consuming manual segmentation. We evaluated the S-UNET on a test-set of 33 images, achieving a median DICE score of 0.88. Apart from high segmentation performance, our method based its decision on two wavelengths with physical meaning for the task-at-hand: 850 nm (peak absorption of oxy-hemoglobin) and 810 nm (isosbestic point of oxy-and deoxy-hemoglobin). Thus, our approach achieves precise data-driven vascular segmentation for automated vascular assessment and may boost MSOT further towards its clinical translation.

bioengineering↗

bioRxiv2019

Domain model explains propagation dynamics and stability of K27 and K36 methylation landscapes

Chromatin states must be maintained during cell proliferation to uphold cellular identity and genome integrity. Inheritance of histone modifications is central in this process. However, the histone modification landscape is challenged by incorporation of new unmodified histones during each cell cycle and the principles governing heritability remain unclear. Here, we take a quantitative computational modeling approach to describes propagation of K27 and K36 methylation states. We measure combinatorial K27 and K36 methylation patterns by quantitative mass spectrometry on subsequent generations of histones. Using model comparison, we reject active global demethylation and invoke the existence of domains defined by distinct methylation endpoints. We find that K27me3 on pre-existing histones stimulates the rate of de novo K27me3 establishment, supporting a read-write mechanism in timely chromatin restoration. Finally, we provide a detailed, quantitative picture of the mutual antagonism between K27 and K37 methylation, and propose that it stabilizes epigenetic states across cell division.

systems biology↗

bioRxiv2019

Gene networks with transcriptional bursting recapitulate rare transient coordinated expression states in cancer

SUMMARYNon-genetic transcriptional variability at the single-cell level is a potential mechanism for therapy resistance in melanoma. Specifically, rare subpopulations of melanoma cells occupy a transient pre-resistant state characterized by coordinated high expression of several genes. Importantly, these rare cells are able to survive drug treatment and develop resistance. How might these extremely rare states arise and disappear within the population? It is unclear whether the canonical stochastic models of probabilistic transcriptional pulsing can explain this behavior, or if it requires special, hitherto unidentified molecular mechanisms. Here we use mathematical modeling to show that a minimal network comprising of transcriptional bursting and interactions between genes can give rise to rare coordinated high states. We next show that although these states occur across networks of different sizes, they depend strongly on three (out of seven) model parameters and require network connectivity to be [≤] 6. Interestingly, we find that while entry into the rare coordinated high state is initiated by a long transcriptional burst that also triggers entry of other genes, the exit from it occurs through the independent inactivation of individual genes. Finally, our model predicts that increased network connectivity can lead to transcriptionally stable states, which we verify using network inference analysis of experimental data. In sum, we demonstrate that established principles of gene regulation are sufficient to describe this new class of rare cell variability and argue for its general existence in other biological contexts.

systems biology↗

bioRxiv2019

Evaluation of domain adaptation approaches for robust classification of heterogeneous biological data sets

Most machine learning algorithms require that training data are identically distributed to ensure effective learning. In biological studies, however, even small variations in the experimental setup can lead to substantial deviations. Domain adaptation offers tools to deal with this problem. It is particularly useful for cases where only a small amount of training data is available in the domain of interest, while a large amount of training data is available in a different, but relevant domain.\n\nWe investigated to what extent domain adaptation was able to improve prediction accuracy for complex biological data. To that end, we used simulated data and time-lapse movies of differentiating blood stem cells in different cell cycle stages from multiple experiments and compared three commonly used domain adaptation approaches. EasyAdapt, a simple technique of structured pooling of related data sets, was able to improve accuracy when classifying the simulated data and cell cycle stages from microscopic images. Meanwhile, the technique proved robust to the potential negative impact on the classification accuracy that is common in other techniques that build models with heterogeneous data. Despite its implementation simplicity, EasyAdapt consistently produced more accurate predictions compared to conventional techniques.\n\nDomain adaptation is therefore able to substantially reduce the amount of work required to create a large amount of annotated training data in the domain of interest necessary whenever the domain changes even a little, which is common not only in biological experiments, but universally exists in almost all data collection routines.

bioinformatics↗

bioRxiv2019

Multi-task learning of a deep k-nearest neighbour network for histopathological image classification and retrieval

Deep neural networks have achieved tremendous success in image recognition, classification and object detection. However, deep learning is often criticised for its lack of transparency and general inability to rationalize its predictions. The issue of poor model interpretability becomes critical in medical applications, as a model that is not understood and trusted by physicians is unlikely to be used in daily clinical practice. In this work, we develop a novel multi-task deep learning framework for simultaneous histopathology image classification and retrieval, leveraging on the classic concept of k-nearest neighbors to improve model interpretability. For a test image, we retrieve the most similar images from our training databases. These retrieved nearest neighbours can be used to classify the test image with a confidence score, and provide a human-interpretable explanation of our classification. Our original framework can be built on top of any existing classification network (and therefore benefit from pretrained models), by (i) adding a triplet loss function with a novel triplet sampling strategy to compare distances between samples and (ii) a Cauchy hashing loss function to accelerate neighbour searching. We evaluate our method on colorectal cancer histology slides, and show that the confidence estimates are strongly correlated with model performance. The explanations provided by nearest neighbors are intuitive and useful for expert evaluation by giving insights into understanding possible model failures, and can support clinical decision making by comparing archived images and patient records with the actual case.

pathology↗

bioRxiv2019

Human-level recognition of blast cells in acute myeloid leukemia with convolutional neural networks

Reliable recognition of malignant white blood cells is a key step in the diagnosis of hematologic malignancies such as Acute Myeloid Leukemia. Microscopic morphological examination of blood cells is usually performed by trained human examiners, making the process tedious, time-consuming and hard to standardise. We compile an annotated image dataset of over 18,000 white blood cells, use it to train a convolutional neural network for leukocyte classification, and evaluate the networks performance. The network classifies the most important cell types with high accuracy. It also allows us to decide two clinically relevant questions with human-level performance, namely (i) if a given cell has blast character, and (ii) if it belongs to the cell types normally present in non-pathological blood smears. Our approach holds the potential to be used as a classification aid for examining much larger numbers of cells in a smear than can usually be done by a human expert. This will allow clinicians to recognize malignant cell populations with lower prevalence at an earlier stage of the disease.

pathology↗

bioRxiv2018

Parameter inference with analytical propagators for stochastic models of autoregulated gene expression

Stochastic gene expression in regulatory networks is conventionally modelled via the Chemical Master Equation (CME). As explicit solutions to the CME, in the form of so-called propagators, are oftentimes not readily available, various approximations have been proposed. A recently developed analytical method is based on a separation of scales that assumes significant differences in the lifetimes of mRNA and protein in the network, allowing for the efficient approximation of propagators from asymptotic expansions for the corresponding generating functions. Here, we showcase the applicability of that method to simulated data from a telegraph model for gene expression that is extended with an autoregulatory mechanism. We demonstrate that the resulting approximate propagators can be applied successfully for Bayesian parameter inference in the non-regulated model; moreover, we show that, in the extended autoregulated model, autoactivation or autorepression may be refuted under certain assumptions on the model parameters. Our results indicate that the method showcased here may allow for successful parameter inference and model identification from longitudinal single cell data.

cell biology↗

bioRxiv2017

Subtle Changes in Clonal Dynamics Underlie the Age-Related Decline in Neurogenesis

Neural stem cells in the adult murine brain have only a limited capacity to self-renew, and the number of neurons they generate drastically declines with age. How cellular dynamics sustain neurogenesis and how alterations with age may result in this decline, are both unresolved issues. Therefore, we clonally traced neural stem cell lineages using confetti reporters in young and old adult mice. To understand underlying mechanisms, we derived mathematical population models of adult neurogenesis that explain the observed clonal cell type abundances. Models fitting the data best consistently show self renewal of transit amplifying progenitors and rapid neuroblast cell cycle exit. Most importantly, we identified an increase of asymmetric stem cell divisions at the expense of symmetric stem cell differentiation with age. Beyond explaining existing longitudinal population data, our model identifies a particular cellular strategy underlying adult neural stem cell homeostasis that gives insights into the aging of a stem cell compartment.

neuroscience↗