Biology subjects

Scherf, N.

Publications and source records attributed to Scherf, N..

9 recordsLinked to original sources

bioRxiv2024

A generalized and versatile framework to train and evaluate autoencoders for biological representation learning and beyond: AUTOENCODIX

Insights and discoveries in complex biological systems, e.g. for personalized medicine, are gained by the combination of large, feature-rich and high-dimensional data with powerful computational methods uncovering patterns and relationships. In recent years, autoencoders, a family of deep learning-based methods for representation learning, are advancing data-driven research due to their variability and non-linear power of multi-modal data integration. Despite their success, current implementations lack standardization, versatility, comparability, and generalizability preventing a broad application. To fill the gap, we present AUTOENCODIX (https://github.com/jan-forest/autoencodix), an open-source framework, designed as a standardized and flexible pipeline for preprocessing, training, and evaluation of autoencoder architectures. These architectures, like ontology-based and cross-modal autoencoders, provide key advantages over traditional methods via explainability of embeddings or the ability to translate across data modalities. We show the value of our framework by its application to data sets from pan-cancer studies (TCGA), single-cell sequencing as well as in combination with imaging. Our studies provide important user-centric insights and recommendations to navigate through architectures, hyperparameters, and important trade-offs in representation learning. Those include reconstruction capability of input data, the quality of embedding for downstream machine learning models, or the reliability of ontology-based embeddings for explainability. In summary, our versatile and generalizable framework allows multi-modal data integration in biomedical research and any other data-driven fields of research. Hence, it can serve as a open-source platform for several major trends and research using autoencoders including architectural improvements, explainability, or training of large-scale pre-trained models.

bioinformatics↗

bioRxiv2024

The utility of explainable A.I. for MRI analysis: Relating model predictions to neuroimaging features of the aging brain

IntroductionDeep learning models highly accurately predict brain-age from MRI but their explanatory capacity is limited. Explainable A.I. (XAI) methods can identify relevant voxels contributing to model estimates, yet, they do not reveal which biological features these voxels represent. In this study, we closed this gap by relating voxel-based contributions to brain-age estimates, extracted with XAI, to human-interpretable structural features of the aging brain. MethodsTo this end, we associated participant-level XAI-based relevance maps extracted from two ensembles of 3D-convolutional neural networks (3D-CNN) that were trained on T1-weighted and fluid attenuated inversion recovery images of 2016 participants (age range 18-82 years), respectively, with regional cortical and subcortical gray matter volume and thickness, perivascular spaces (PVS) and water diffusion-based fractional anisotropy of main white matter tracts. ResultsWe found that all neuroimaging markers of brain aging, except for PVS, were highly correlated with the XAI-based relevance maps. Overall, the strongest correlation was found between ventricular volume and relevance (r = 0.69), and by feature, temporal-parietal cortical thickness and volume, cerebellar gray matter volume and frontal-occipital white matter tracts showed the strongest correlations with XAI-based relevance. ConclusionOur ensembles of 3D-CNNs took into account a plethora of known aging processes in the brain to perform age prediction. Some age-associated features like PVS were not consistently considered by the models, and the cerebellum was more important than expected. Taken together, we highlight the ability of end-to-end deep learning models combined with XAI to reveal biologically relevant, multi-feature relationships in the brain.

neuroscience↗

bioRxiv2024

Benchmarking Large Language Models for Bio-Image Analysis Code Generation

In the computational age, life-scientists often have to write Python code to solve bio-image analysis (BIA) problems. Many of them have not been formally trained in programming though. Code-generation, or coding assistance in general, with Large Language Models (LLMs) can have a clear impact on BIA. To the best of our knowledge, the quality of the generated code in this domain has not been studied. We present a quantitative benchmark to estimate the capability of LLMs to generate code for solving common BIA tasks. Our benchmark currently consists of 57 human-written prompts with corresponding reference solutions in Python, and unit-tests to evaluate functional correctness of potential solutions. We demonstrate our benchmark here and compare 18 state-of-the-art LLMs. To ensure that we will cover most of our community needs we also outline mid- and long-term strategies to maintain and extend the benchmark by the BIA open-source community. This work should support users in deciding for an LLM and also guide LLM developers in improving the capabilities of LLMs in the BIA domain.

bioinformatics↗

bioRxiv2020

On the Ontological Foundations of Cellular Development

Time-lapse microscopy is a principal tool to unravel the mystery of how cells form and maintain organisms. The complexity of the domain of cellular dynamics demands a conceptual architecture as a solid theoretical foundation that supports the integration of knowledge obtained across experiments and theories. In this work, we outline the ontological foundation of cellular genealogies, a key concept for describing and representing of cellular development. We build the conceptual framework following the onto-axiomatic method: We first analyse the domain within the context of a top-level ontology (GFO). The resulting domain-specification provides the basis for a conceptualisation where we introduce concepts and relations. From these conceptualisations, we then construct model-structures adhering to the principles of model-theory. We finally elaborate axioms based on these model-structures. The developed framework provides the fundamental concepts underlying a Cell Tracking Ontology (CTO) that supports extraction and integration of biological knowledge from systems-level experiments across different types of observations at the single-cell level.

bioinformatics↗

bioRxiv2020

linus: Conveniently explore, share and present large-scale biological trajectory data from a web browser.

In biology, we are often confronted with information-rich, large-scale trajectory data, but exploring and communicating patterns in such data is often a cumbersome task. Ideally, the data should be wrapped with an interactive visualisation in one concise package that makes it straightforward to create and test hypotheses collaboratively. To address these challenges, we have developed a tool, linus, which makes the process of exploring and sharing 3D trajectories as easy as browsing a website. We provide a python script that reads trajectory data and enriches them with additional features, such as edge bundling or custom axes and generates an interactive web-based visualisation that can be shared offline and online. The goal of linus is to facilitate the collaborative discovery of patterns in complex trajectory data.

bioinformatics↗

bioRxiv2020

TraCurate: efficiently curating cell tracks

TraCurate is an open-source software tool to curate and manually annotate cell tracking data from time-lapse microscopy. Although many studies of cellular behavior require high-quality, long-term observations across generations of cells, automated cell tracking is often imperfect and typically yields fragmented results that still contain many errors. TraCurate provides the functionality for the curation and correction of cell tracking data with minimal user interaction and expenditure of time and supports the extraction of complete cell tracks and cellular genealogies from experimental data. Source code and binary packages for Linux, macOS and Windows are available at https://tracurate.gitlab.io/, as well as all other complementary tools described herein.

bioinformatics↗

bioRxiv2018

Deep Learning meets Topology-preserving Active Contours: towards scalable quantitative histology of cortical cytoarchitecture.

Deep learning has thoroughly changed the field of image analysis yielding impressive results whenever enough annotated data can be gathered. While partial annotation can be very fast, manual segmentation of 3D biological structures is tedious and error-prone. Additionally, high-level shape concepts such as topology or boundary smoothness are hard if not impossible to encode in Feedforward Neural Networks. Here we present a modular strategy for the accurate segmentation of neural cell bodies from light-sheet microscopy combining mixed-scale convolutional neural networks and topology-preserving geometric deformable models. We show that the network can be trained efficiently from simple cell centroid annotations, and that the final segmentation provides accurate cell detection and smooth segmentations that do not introduce further cell splitting or merging.

bioinformatics↗

bioRxiv2017

Pan-embryo cell dynamics of germlayer formation in zebrafish

Cell movements are coordinated across spatio-temporal scales to achieve precise positioning of organs during vertebrate gastrulation. In zebrafish, mechanisms governing such morphogenetic movements have so far only been studied within a local region or a single germlayer. Here, we present pan-embryo analyses of fate specification and dynamics of all three germlayers simultaneously within a gastrulating embryo, showing that cell movement characteristics are predominantly determined by its position within the embryo, independent of its germlayer identity. The spatially confined fate specification establishes a distinct distribution of cells in each germlayer during early gastrulation. The differences in the initial distribution are subsequently amplified by a unique global movement, which organizes the organ precursors along the embryonic body axis, giving rise to the blueprint of organ formation.

developmental biology↗

bioRxiv2017

Cell-Accurate Optical Mapping Across The Entire Developing Heart

Organogenesis depends on orchestrated interactions between individual cells and morphogenically relevant cues at the tissue level. This is true for the heart, whose function critically relies on well-ordered communication between neighbouring cells, which is established and fine-tuned during development. For an integrated understanding of the development of structure and function, we need to move from isolated snap-shot observations of either microscopic or macroscopic parameters to simultaneous and, ideally continuous, cell-to-organ scale imaging. We introduce cell-accurate three-dimensional Ca2+-mapping of all cells in the entire heart during the looping stage in live embryonic zebrafish, using high-speed light sheet microscopy and tailored image processing and analysis. We show how myocardial region-specific heterogeneity in cell function emerges during early development and how structural patterning goes hand-in-hand with functional maturation of the entire heart. Our method opens the way to systematic, scale-bridging, in vivo studies of vertebrate organogenesis by cell-accurate structure-function mapping across entire organs.

biophysics↗