Biology subjects

Hauschild, A.-C.

Publications and source records attributed to Hauschild, A.-C..

10 recordsLinked to original sources

bioRxiv2026

Benchmarking Graph Neural Networks for Multi-Omics Cancer Subtyping using Methylation and Gene Expression Profiles

Motivation: Graph Neural Networks (GNNs) have gained increasing interest in the biomedical domain, as the integration of prior knowledge and deep neural networks has the potential to enhance insights into molecular processes and disease mechanisms. However, a comprehensive and systematic assessment of model architectures, data modalities, graph structures, and their performance for graph signal classification in the biomedical domain is yet to be performed. In order to close this gap, we conducted a benchmarking study on multiple GNNs on a Protein-Protein Interaction (PPI) network for Kidney Renal Clear Cell Carcinoma and Breast cancer subtype prediction, performing an in-depth investigation of architectures, incorporating skip connections and various data modalities. Results: While none of the GNNs outperforms the structure-agnostic Multi-Layer Perceptron baseline, all of them can handle bimodal data (gene methylation and expression) and offer the ability to gain explainability based on PPIs. We offer practical guidelines for applying GNNs to graph signal processing tasks specifically for cancer classification. Depending on the underlying dataset and PPI structure employed, models on different data modalities outperform others. Overall, we suggest using ChebNet, which tends to outperform the Graph Convolutional Network and the Graph Attention Network in cancer subtype prediction. We recommend using GNN architectures that employ a simple flattening readout layer, as they provide better classification performance and faster training time than those with global average pooling. Additionally, we tested residual connections, but they had only an insignificant impact on classification performance.

bioinformatics↗

bioRxiv2025

Ensemble Post-hoc Explainable AI in Multivariate Time Series: Identifying Medical Features Driving Disease Prediction

BackgroundDespite the growing success of deep learning (DL) in multivariate time-series classification, such as 12-lead electro-cardiography (ECG), widespread integration into clinical practice has yet to be achieved. The limited transparency of DL hinders clinical adoption, where understanding model decisions is crucial for trust and compliance with regulations such as the General Data Protection Regulation (GDPR) or the EU AI Act. ResultsTo tackle this challenge, we implemented a widely used 1D-ResNet in Pytorch that was trained on the large-scale Brazilian CODE dataset to classify six different ECG abnormalities. We employed the model on the German PTB XL dataset, and evaluated its decision-making processes using 16 post-hoc explainable AI (XAI) methods. To assess the clinical relevance of the models attributions, we conducted a Wilcoxon signed-rank test to identify features with significantly higher relevance for each XAI method. We used an ensemble majority vote approach to validate whether the model has learned clinically meaningful features for each abnormality. Additionally, a Mann-Whitney U test was employed to detect significant differences in relevance attributions between correctly and incorrectly classified ECGs. Overall, the model achieved sensitivity scores above 0.9 for most abnormalities in the PTB XL dataset. However, our XAI analysis showed that the model struggled to capture clinically relevant features for some diseases. Certain XAI methods, including DeepLift, DeepLiftShap, and Occlusion, consistently highlighted clinically meaningful features across abnormalities, while others, such as LIME, KernelShap, and LRP, failed to do so. Moreover, some XAI methods demonstrated significant differences in attributions between correctly and incorrectly classified ECGs, highlighting their potential for enhancing model robustness and interpretability. ConclusionOur findings underscore the importance of selecting suitable XAI methods tailored to specific model architectures and data types to ensure transparency and reliability. By identifying effective XAI techniques, this study contributes to closing the gap between DL advancements and their clinical implementation, paving the way for more trustworthy AI-driven healthcare solutions.

bioinformatics↗

bioRxiv2024

BenchXAI: Comprehensive Benchmarking of Post-hoc Explainable AI Methods on Multi-Modal Biomedical Data

The increasing digitalisation of multi-modal data in medicine and novel artificial intelligence (AI) algorithms opens up a large number of opportunities for predictive models. In particular, deep learning models show great performance in the medical field. A major limitation of such powerful but complex models originates from their black-box nature. Recently, a variety of explainable AI (XAI) methods have been introduced to address this lack of transparency and trust in medical AI. However, the majority of such methods have solely been evaluated on single data modalities. Meanwhile, with the increasing number of XAI methods, integrative XAI frameworks and benchmarks are essential to compare their performance on different tasks. For that reason, we developed BenchXAI, a novel XAI benchmarking package supporting comprehensive evaluation of fifteen XAI methods, investigating their robustness, suitability, and limitations in biomedical data. We employed BenchXAI to validate these methods in three common biomedical tasks, namely clinical data, medical image and signal data, and biomolecular data. Our newly designed sample-wise normalisation approach for post-hoc XAI methods enables the statistical evaluation and visualisation of performance and robustness. We found that the XAI methods Integrated Gradients, DeepLift, DeepLiftShap, and GradientShap performed well over all three tasks, while methods like Deconvolution, Guided Backpropagation, and LRP-1-{beta}0 struggled for some tasks. With acts such as the EU AI Act the application of XAI in the biomedical domain becomes more and more essential. Our evaluation study represents a first step toward verifying the suitability of different XAI methods for various medical domains.

bioinformatics↗

bioRxiv2024

Effect of Data Heterogeneity in Clinical MALDI-TOF Mass Spectra Profiles on Direct Antimicrobial Resistance Prediction through Machine Learning

The matrix-assisted laser desorption-ionization time-of-flight mass spectrometry has become a powerful tool for accurate species identification in routine diagnostic microbiology. Recently, the application of machine learning models with MALDI-TOF mass spectra data indicated that rapid prediction of antimicrobial resistance patterns might facilitate even timelier and improved antimicrobial treatment. Although MALDI-TOF mass spectra data have proven valuable for clinical decision support, the issue of class imbalance in routine clinical data is often overlooked. This imbalance arises from factors such as local epidemiology, selective pressure from antibiotics, culture conditions, the methodology of phenotypic antimicrobial susceptibility testing, and sample preparation processes. Here, we provide a large mass spectra dataset, MS-UMG, for antimicrobial resistance prediction model training. With previously available public datasets, our dataset is evaluated and validated for usage in AMR prediction. We further explore the mass spectra data and identify informative regions on the spectra profile for AMR prediction. Moreover, we investigate the composition of this clinical dataset and present the implications of data heterogeneity on machine learning model performance. In conclusion, our findings highlight that accurate comprehension of clinical routine data and consideration of diverse hospital protocols are critical for effective clinical decision support systems with machine learning models. Key PointsO_LIIntroduced a large-scale clinical mass spectrometry dataset to the scientific community for research on antimicrobial resistance. C_LIO_LIConducted a comparison and evaluation of this dataset with other existing large-scale MS datasets, highlighting its value for developing and validating predictive models in clinical settings. C_LIO_LIDemonstrated the robustness of machine learning models for antimicrobial resistance prediction using large-scale clinical mass spectra profiles. C_LIO_LIAnalyzed the impact of data heterogeneity on the training and performance of machine learning models, emphasizing the need to account for variability in clinical routine data to enhance model reliability and generalizability. C_LI

bioinformatics↗

bioRxiv2024

CODEX: COunterfactual Deep learning for the in-silico EXploration of cancer cell line perturbations

MotivationHigh-throughput screens (HTS) provide a powerful tool to decipher the causal effects of chemical and genetic perturbations on cancer cell lines. Their ability to evaluate a wide spectrum of interventions, from single drugs to intricate drug combinations and CRISPR-interference, has established them as an invaluable resource for the development of novel therapeutic approaches. Nevertheless, the combinatorial complexity of potential interventions makes a comprehensive exploration intractable. Hence, prioritizing interventions for further experimental investigation becomes of utmost importance. ResultsWe propose CODEX as a general framework for the causal modeling of HTS data, linking perturbations to their downstream consequences. CODEX relies on a stringent causal modeling strategy based on counterfactual reasoning. As such, CODEX predicts drug-specific cellular responses, comprising cell survival and molecular alterations, and facilitates the in-silico exploration of drug combinations. This is achieved for both bulk and single-cell HTS. We further show that CODEX provides a rationale to explore complex genetic modifications from CRISPR-interference in silico in single cells. Availability and ImplementationOur implementation of CODEX is publicly available at https://github.com/sschrod/CODEX. All data used in this article are publicly available.

bioinformatics↗

bioRxiv2023

Species-Agnostic Transfer Learning for Cross-species Transcriptomics Data Integration without Gene Orthology

Novel hypotheses in biomedical research are often developed or validated in model organisms such as mice and zebrafish and thus play a crucial role, particularly in studying disease mechanisms and treatment responses. However, due to biological differences between species, translating these findings into human applications remains challenging. Moreover, commonly used orthologous gene information is often incomplete, particularly for non-model organisms, and entails a significant information loss during gene-id conversion. To address these issues, we present a novel methodology for species-agnostic transfer learning with heterogeneous domain adaptation. We built on the cross-domain structure-preserving projection and extended the algorithm toward out-of-sample prediction, a common challenge in biomedical sequencing data. Our approach not only allows knowledge integration and translation across various species without relying on gene orthology but also identifies similar GO biological processes amongst the most influential genes composing the latent space for species integration. Subsequently, this enables the identification and functional annotation of genes missing from public orthology databases. Finally, we evaluated our approach with four different single-cell sequencing datasets focusing on out-of-sample prediction and compared it against related machine-learning approaches. In summary, the developed model outperforms all related methods working without prior knowledge when predicting unseen cell types based on other species data. The results demonstrate that our novel approach allows knowledge transfer beyond species barriers without the dependency on known gene orthology but utilizing the entire gene sets.

bioinformatics↗

bioRxiv2023

Ensemble-GNN: federated ensemble learning with graph neural networks for disease module discovery and classification

Federated learning enables collaboration in medicine, where data is scattered across multiple centers without the need to aggregate the data in a central cloud. While, in general, machine learning models can be applied to a wide range of data types, graph neural networks (GNNs) are particularly developed for graphs, which are very common in the biomedical domain. For instance, a patient can be represented by a protein-protein interaction (PPI) network where the nodes contain the patient-specific omics features. Here, we present our Ensemble-GNN software package, which can be used to deploy federated, ensemble-based GNNs in Python. Ensemble-GNN allows to quickly build predictive models utilizing PPI networks consisting of various node features such as gene expression and/or DNA methylation. We exemplary show the results from a public dataset of 981 patients and 8469 genes from the Cancer Genome Atlas (TCGA).

bioinformatics↗

bioRxiv2022

Interactive explainable AI platform for graph neural networks

BackgroundLack of trust in artificial intelligence (AI) models in medicine is still the key blockage for the use of AI in clinical decision support systems (CDSS). Although AI models are already performing excellently in systems medicine, their black-box nature entails that patient-specific decisions are incomprehensible for the physician. This is especially true for very complex models such as graph neural networks (GNNs), a common state-of-the-art approach to model biological networks such as protein-protein-interaction graphs (PPIs) to predict clinical outcomes. The aim of explainable AI (XAI) algorithms is to "explain" to a human domain expert, which input features, such as genes, influenced a specific recommendation. However, in the clinical domain, it is essential that these explanations lead to some degree of causal understanding by a clinician in the context of a specific application. ResultsWe developed the CLARUS platform, aiming to promote human understanding of GNN predictions by allowing the domain expert to validate and improve the decision-making process. CLARUS enables the visualisation of the patient-specific biological networks used to train and test the GNN model, where nodes and edges correspond to gene products and their interactions, for instance. XAI methods, such as GNNExplainer, compute relevance values for genes and interactions. The CLARUS graph visualisation highlights gene and interaction relevances by color intensity and line thickness, respectively. This enables domain experts to gain deeper insights into the biological network by identifying the most influential sub-graphs and molecular pathways crucial for the decision-making process. More importantly, the expert can interactively alter the patient-specific PPI network based on the acquired understanding and initiate re-prediction or retraining. This interactivity allows to ask manual counterfactual questions and analyse the resulting effects on the GNN prediction. ConclusionTo the best of our knowledge, we present the first interactive XAI platform prototype, CLARUS, that allows not only the evaluation of specific human counterfactual questions based on user-defined alterations of patient PPI networks and a re-prediction of the clinical outcome but also a retraining of the entire GNN after changing the underlying graph structures. The platform is currently hosted by the GWDG on https://rshiny.gwdg.de/apps/clarus/.

bioinformatics↗

bioRxiv2022

On the importance of data transformation for data integration in single-cell RNA sequencing analysis

AO_SCPLOWBSTRACTC_SCPLOWRecent advances in single-cell RNA (scRNA) sequencing have opened a multitude of possibilities to study tissues down to the level of cellular populations. Subsequently, this enabled various scRNA studies that reported novel or previously undetected subpopulations and their functions by integrating multiple datasets. However, the heterogeneity in single-cell sequencing data makes it unfeasible to adequately integrate multiple datasets generated from different studies. This heterogeneity originates from various sources of noise due to technological limitations. Thus, particular procedures are required to adjust such effects prior to further integrative analysis. Over the last years, numerous single-cell data analysis workflows have been introduced, implementing various read-count transformation methods for de-noising and batch correction. A detailed review of recent single-cell studies shows while many analysis procedures employ various preprocessing steps, they often neglect the importance of a well-chosen and optimized data transformation. This fact is particularly alarming since these data transformations can alter data distribution and thus have a crucial impact on subsequent downstream cell clustering results. Therefore, this study investigates the effects of the various data transformation methods on three different public data scenarios and evaluates them with the most commonly used dimensionality reduction and clustering analysis. Additionally, we discuss its implications for the subsequent application of different deep neural network approaches, such as auto encoders and transfer learning. In summary, our benchmark analysis shows that a large portion of batch effects and noise can be mitigated by simple but well-chosen data transformation methods. We conclude that such optimized preprocessing is crucial and should be the baseline for all comparative single-cell sequencing studies, particularely for integrative analysis of multiple data sets.

bioinformatics↗

bioRxiv2021

Transfer Learning Compensates Limited Data, Batch-Effects, And Technical Heterogeneity In Single-Cell Sequencing

AO_SCPLOWBSTRACTC_SCPLOWTremendous advances in next-generation sequencing technology have enabled the accumulation of large amounts of omics data in various research areas over the past decade. However, study limitations due to small sample sizes, especially in rare disease clinical research, technological heterogeneity, and batch effects limit the applicability of traditional statistics and machine learning analysis. Here, we present a meta-learning approach to transfer knowledge from big data and reduce the search space in data with small sample sizes. Few-shot learning algorithms integrate meta-learning to overcome data scarcity and data heterogeneity by transferring molecular pattern recognition models from datasets of unrelated domains. We explore few-shot learning models with large scale public dataset, TCGA (The Cancer Genome Atlas) and GTEx dataset, and demonstrate their potential as meta-learning dataset in other molecular pattern recognition tasks. Our results show that transfer learning is very effective for datasets with a limited sample size. Furthermore, we show that our approach can transfer knowledge across technological heterogeneity, e.g., from bulk cell to single-cell data. Our approach can overcome study size constraints, batch effects, and technological limitations in analyzing single-cell data by leveraging existing bulk-cell sequencing data.

bioinformatics↗