Biology subjects

Wurmus, R.

Publications and source records attributed to Wurmus, R..

6 recordsLinked to original sources

bioRxiv2024

Enhancing Biomarker-Based Oncology Trial Matching Using Large Language Models

Clinical trials are an essential component of drug development for new cancer treatments, yet the information required to determine a patients eligibility for enrollment is scattered in large amounts of unstructured text. Genomic biomarkers are especially important in precision medicine and targeted therapies, making them essential for matching patients to appropriate trials. Large language models (LLMs) offer a promising solution for extracting this information from clinical trial data, aiding both physicians and patients in identifying suitable matches. In this study, we explore various LLM strategies for extracting genetic biomarkers from oncology trials to improve patient enrollment rates. Our results show that open-source language models, when applied out-of-the-box, effectively capture complex logical expressions and structure genomic biomarkers in disjunctive normal form, outperforming closed-source models such as GPT-4 and GPT-3.5-Turbo. Furthermore, fine-tuning these open-source models with additional data significantly enhances their performance.

bioinformatics↗

bioRxiv2024

Flexynesis: A deep learning framework for bulk multi-omics data integration for precision oncology and beyond

Accurate decision making in precision oncology depends on integration of multimodal molecular information, such as the genetic data, gene expression, protein abundance, and epigenetic measurements. Deep learning methods facilitate integration of heterogeneous datasets. However, almost all published deep learning-based bulk multi-omics integration methods have constrained usability. They suffer from lack of transparency, modularity, deployability, and are applicable exclusively to narrow tasks. To address these limitations, we introduce Flexynesis, a versatile tool designed with usability, and adaptability in mind. Flexynesis streamlines data processing, enforces structured data splitting, and ensures rigorous model evaluation. It offers unsupervised feature selection, different omics layer fusion options, and hyperparameter tuning. Users can choose from distinct architectures - fully connected networks, variational autoencoders, multi-triplet networks, graph neural networks, and cross-modality encoding networks. Each model is complemented with a straightforward input interface and standardized training, evaluation, and feature importance quantification methods, enabling easy incorporation into data integration pipelines. For improved user experience, Flexynesis supports features such as on-the-fly task determination and compatibility with regression, classification, and survival modeling. It accommodates multi-task prediction of a mixture of numerical/categorical outcome variables with a tolerance for missing labels. We also developed an extensive benchmarking pipeline, showcasing the tools capability across diverse real-life datasets. This toolset should make deep-learning based bulk multi-omics data integration in the context of clinical/pre-clinical data analysis and marker discovery more accessible to a wider audience with or without experience in deep-learning development. Flexynesis is available at https://github.com/BIMSBbioinfo/flexynesis and can be installed from https://pypi.org/project/flexynesis/. O_FIG O_LINKSMALLFIG WIDTH=199 HEIGHT=200 SRC="FIGDIR/small/603606v1_ufig1.gif" ALT="Figure 1"> View larger version (53K): org.highwire.dtl.DTLVardef@17eda85org.highwire.dtl.DTLVardef@13c89ecorg.highwire.dtl.DTLVardef@182f706org.highwire.dtl.DTLVardef@127d9d9_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical Abstract:C_FLOATNO Summary of the Flexynesis data integration and analysis workflow. C_FIG

bioinformatics↗

bioRxiv2021

Optogenetic perturbations of RNA expression in tissue space

Organoids derived from stem cells become increasingly important to study human development and to model disease. However, methods are needed to control and study spatio-temporal patterns of gene expression in organoids. To this aim, we combined optogenetics and gene perturbation technologies to activate or knock-down RNA of target genes, at single-cell resolution and in programmable spatio-temporal patterns. To illustrate the usefulness of our approach, we locally activated Sonic Hedgehog (SHH) signaling in an organoid model for human neurodevelopment. High-resolution spatial transcriptomic and single-cell analyses showed that this local induction was sufficient to generate stereotypically patterned organoids in three dimensions and revealed new insights into SHHs contribution to gene regulation in neurodevelopment. With this study, we propose optogenetic perturbations in combination with spatial transcriptomics as a powerful technology to reprogram and study cell fates and tissue patterning in organoids.

bioengineering↗

bioRxiv2018

Reproducible genomics analysis pipelines with GNU Guix

In bioinformatics, as well as other computationally-intensive research fields, there is a need for workflows that can reliably produce consistent output, independent of the software environment or configuration settings of the machine on which they are executed. Indeed, this is essential for controlled comparison between different observations or for the wider dissemination of workflows. Providing this type of reproducibility, however, is often complicated by the need to accommodate the myriad dependencies included in a larger body of software, each of which generally come in various versions. Moreover, in many fields (bioinformatics being a prime example), these versions are subject to continual change due to rapidly evolving technologies, further complicating problems related to reproducibility. Here, we propose a principled approach for building analysis pipelines and managing their dependencies. As a case study to demonstrate the utility of our approach, we present a set of highly reproducible pipelines for the analysis of RNA-seq, ChIP-seq, Bisulfite-seq, and single-cell RNA-seq. All pipelines process raw experimental data, and generate reports containing publication-ready plots and figures, with interactive report elements and standard observables. Users may install these highly reproducible packages and apply them to their own datasets without any special computational expertise beyond the use of the command line. We hope such a toolkit will provide immediate benefit to laboratory workers wishing to process their own data sets or bioinformaticians seeking to automate all, or parts of, their analyses. In the long term, we hope our approach to reproducibility will serve as a blueprint for reproducible workflows in other areas. Our pipelines, along with their corresponding documentation and sample reports, are available at http://bioinformatics.mdc-berlin.de/pigx

bioinformatics↗

bioRxiv2018

Reproducible inference of transcription factor footprints in ATAC-seq and DNase-seq datasets via protocol-specific bias modeling

DNase-seq and ATAC-seq are broadly used methods to assay open chromatin regions genome-wide. The single nucleotide resolution of DNase-seq has been further exploited to infer transcription factor binding sites (TFBS) in regulatory regions via footprinting. Recent studies have demonstrated the sequence bias of DNase I and its adverse effects on footprinting efficiency. However, footprinting and the impact of sequence bias have not been extensively studied for ATAC-seq. Here, we undertake a systematic comparison of the two methods and show that a modification to the ATAC-seq protocol increases its yield and its agreement with DNase-seq data from the same cell line. We demonstrate that the two methods have distinct sequence biases and correct for these protocol-specific biases when performing footprinting. Despite differences in footprint shapes, the locations of the inferred footprints in ATAC-seq and DNase-seq are largely concordant. However, the protocol-specific sequence biases in conjunction with the sequence content of TFBSs impacts the discrimination of footprint from background, which leads to one method outperforming the other for some TFs. Finally, we address the depth required for reproducible identification of open chromatin regions and TF footprints.

genomics↗

bioRxiv2017

HOT or not: Examining the basis of high-occupancy target regions

High-occupancy target (HOT) regions are the segments of the genome with unusually high number of transcription factor binding sites. These regions are observed in multiple species and thought to have biological importance due to high transcription factor occupancy. Furthermore, they coincide with house-keeping gene promoters and the associated genes are stably expressed across multiple cell types. Despite these features, HOT regions are solemnly defined using ChIP-seq experiments and shown to lack canonical motifs for transcription factors that are thought to be bound there. Although, ChIP-seq experiments are the golden standard for finding genome-wide binding sites of a protein, they are not noise free. Here, we show that HOT regions are likely to be ChIP-seq artifacts and they are similar to previously proposed \"hyper-ChIPable\" regions. Using ChIP-seq data sets for knocked-out transcription factors, we demonstrate presence of false positive signals on HOT regions. We observe sequence characteristics and genomic features that are discriminatory of HOT regions, such as GC/CpG-rich k-mers and enrichment of RNA-DNA hybrids (R-loops) and DNA tertiary structures (G-quadruplex DNA). The artificial ChIP-seq enrichment on HOT regions could be associated to these discriminatory features. Furthermore, we propose strategies to deal with such artifacts for the future ChIP-seq studies.

genomics↗