Biology subjects

Schuppert, A.

Publications and source records attributed to Schuppert, A..

5 recordsLinked to original sources

bioRxiv2021

Swarm learning for decentralized artificial intelligence in cancer histopathology

Artificial Intelligence (AI) can extract clinically actionable information from medical image data. In cancer histopathology, AI can be used to predict the presence of molecular alterations directly from routine histopathology slides. However, training robust AI systems requires large datasets whose collection faces practical, ethical and legal obstacles. These obstacles could be overcome with swarm learning (SL) where partners jointly train AI models, while avoiding data transfer and monopolistic data governance. Here, for the first time, we demonstrate the successful use of SL in large, multicentric datasets of gigapixel histopathology images comprising over 5000 patients. We show that AI models trained using Swarm Learning can predict BRAF mutational status and microsatellite instability (MSI) directly from hematoxylin and eosin (H&E)-stained pathology slides of colorectal cancer (CRC). We trained AI models on three patient cohorts from Northern Ireland, Germany and the United States of America and validated the prediction performance in two independent datasets from the United Kingdom using SL-based AI models. Our data show that SL enables us to train AI models which outperform most locally trained models and perform on par with models which are centrally trained on the merged datasets. In addition, we show that SL-based AI models are data efficient and maintain a robust performance even if only subsets of local datasets are used for training. In the future, SL can be used to train distributed AI models for any histopathology image analysis tasks, overcoming the need for data transfer and without requiring institutions to give up control of the final AI model.

bioinformatics↗

bioRxiv2020

Plant PhysioSpace: a robust tool to compare stress response across plant species

Generalization of transcriptomics results can be achieved by comparison across experiments, which is based on integration of interrelated transcriptomics studies into a compendium. Both characterization of the fate of the organism as well as distinguishing between generic and specific responses can be gained in such a broader context. There are numerous methods for analyzing such data sets, most focusing on gene-wise dimension reduction to obtain marker genes and gene sets, e.g. for pathway analysis. Relying only on isolated biological modules might lead to missing of important confounders and relevant context. We have developed a novel method called Plant PhysioSpace, which provides the ability to compute experimental conditions across species and platforms without a priori reducing the reference information to specific gene-sets. It extracts physiologically relevant signatures from a reference data set, a collection of public data sets, by integrating and transforming heterogeneous reference gene expression data into a set of physiology-specific patterns. New experimental data can be mapped to these patterns, resulting in similarity scores which provide quantitative likeness of the new experiment to the a priori compendium. Because of its robustness against noise and platform bias, Plant PhysioSpace can be used as an inter-species or cross-platform similarity measure. We have demonstrated its success in translating stress responses between different species and platforms (including single-cell technologies). We also report the implementation of two R packages, one software and one data package, and a shiny web application, which provides plant biologists convenient ways to access our method and precomputed models from four different species.

plant biology↗

bioRxiv2020

Two-step multi-omics modelling of drug sensitivity in cancer cell lines to identify driving mechanisms

Drug sensitivity prediction models for human cancer cell lines constitute important tools in identifying potential driving factors of responsiveness in a pre-clinical setting. Integrating information derived from a range of heterogeneous data is crucial, but remains non-trivial, as differences in data structures may hinder fitting algorithms from assigning adequate weights to complementary information that is contained in distinct omics data. In order to counteract this effect that tends to lead to just one data type dominating supposedly multi-omics models, we developed a novel tool that enables users to train single-omics models separately in a first step and to integrate them into a multi-omics model in a second step. Extensive ablation studies are performed in order to facilitate an in-depth evaluation of the respective contributions of singular data types and of combinations thereof, effectively identifying redundancies and interdependencies between them. Moreover, the integration of the single-omics models is realized by a range of distinct classification algorithms, thus allowing for a performance comparison. Sets of molecular events and tissue types found to be related to significant shifts in drug sensitivity are returned to facilitate a comprehensive and straightforward analysis of potential drivers of drug responsiveness. Our two-step approach yields sets of actual multi-omics pan-cancer classification models that are highly predictive for a majority of drugs in the GDSC data base. In the context of targeted drugs with particular modes of action, its predictive performances compare favourably to those of classification models that incorporate multi-omics data in a simple one-step approach. Additionally, case studies demonstrate that it succeeds both in correctly identifying known key drivers of specific drug compounds as well as in providing sets of potential candidates for additional driving factors of drug sensitivity.

systems biology↗

bioRxiv2017

A Systematic Atlas of Chaperome Deregulation Topologies Across the Human Cancer Landscape

Proteome balance is safeguarded by the proteostasis network (PN), an intricately regulated network of conserved processes that evolved to maintain native function of the diverse ensemble of protein species, ensuring cellular and organismal health. Proteostasis imbalances and collapse are implicated in a spectrum of human diseases, from neurodegeneration to cancer. The characteristics of PN disease alterations however have not been assessed in a systematic way. Zooming in on the chaperome as a central PN component we turned to a curated functional ontology of the human chaperome that we connect in a high-confidence physical protein-protein interaction network. Challenged by the lack of a systems-level understanding of proteostasis alterations in the heterogeneous spectrum of human cancers, we assessed gene expression across more than 10,000 patient biopsies covering 22 solid cancers. We derived a novel customized Meta-PCA dimension reduction approach yielding M-scores as quantitative indicators of disease expression changes to condense the complexity of cancer transcriptomics datasets into quantitative functional network topographies. We confirm upregulation of the HSP90 family and also highlight HSP60s, Prefoldins, HSP100s, ER- and mitochondria-specific chaperones as pan-cancer enriched. Our analysis also reveals a surprisingly consistent strong downregulation of small heat shock proteins (sHSPs) and we stratify two cancer groups based on the preferential upregulation of ATP-dependent chaperones. Strikingly, our analysis highlight similarities between stem cell and cancer proteostasis, and diametrically opposed chaperome deregulation between cancers and neurodegenerative diseases. We developed a web-based Proteostasis Profiler tool (Pro2) enabling intuitive analysis and visual exploration of proteostasis disease alterations using gene expression data. Our study showcases a comprehensive profiling of chaperome shifts in human cancers and sets the stage for a systematic global analysis of PN alterations across the human diseasome towards novel hypotheses for therapeutic network re-adjustment in proteostasis disorders.

systems biology↗

bioRxiv2017

Stem cell differentiation is a stochastic process with memory

Pluripotent stem cells are able to self-renew indefinitely in culture and differentiate into all somatic cell types in vivo. While much is known about the molecular basis of pluripotency, the molecular mechanisms of lineage commitment are complex and only partially understood. Here, using a combination of single cell profiling and mathematical modeling, we examine the differentiation dynamics of individual mouse embryonic stem cells (ESCs) as they progress from the ground state of pluripotency along the neuronal lineage. In accordance with previous reports we find that cells do not transit directly from the pluripotent state to the neuronal state, but rather first stochastically permeate an intermediate primed pluripotent state, similar to that found in the maturing epiblast in development. However, analysis of rate at which individual cells enter and exit this intermediate metastable state using a hidden Markov model reveals that the observed ESC and epiblast-like macrostates conceal a chain of unobserved cellular microstates, which individual cells transit through stochastically in sequence. These hidden microstates ensure that individual cells spend well-defined periods of time in each functional macrostate and encode a simple form of epigenetic memory that allows individual cells to record their position on the differentiation trajectory. To examine the generality of this model we also consider the differentiation of mouse hematopoietic stem cells along the myeloid lineage and observe remarkably similar dynamics, suggesting a general underlying process. Based upon these results we suggest a statistical mechanics view of cellular identities that distinguishes between functionally-distinct macrostates and the many functionally-similar molecular microstates associated with each macrostate. Taken together these results indicate that differentiation is a discrete stochastic process amenable to analysis using the tools of statistical mechanics.

systems biology↗