Rolf Backofen

StoatyDive evaluates and classifies peak profiles from sequencing data to improve downstream analysis and quality control.

GraphProt2 predicts RNA-binding protein binding sites using graph neural networks, outperforming existing methods on CLIP-seq datasets.

The paper develops a deep learning model to recommend Galaxy tools based on workflow analysis, achieving 97% accuracy in tool prediction.

CRISPRIdentify uses machine learning to detect CRISPR arrays with reduced false positives and provides certainty scores for identified regions.

HRIBO is a workflow for analyzing bacterial Ribo-seq data to discover novel ORFs, using annotation-independent methods and integrating additional features for prioritization.

The paper introduces a Galaxy-based suite for scRNA-seq analysis, offering workflows, training resources, and tools for preprocessing, normalization, and clustering across 10x and alternative platforms.

The paper introduces a machine learning approach to identify and classify CRISPR-Cas systems by combining regression and classification models to improve protein cascade quality and subtype prediction.

The paper introduces an approximate k-nearest neighbors method using MinHash to reduce the dimensionality of single-cell Hi-C data, enabling efficient clustering of large datasets.

GraphClust2 improves RNA clustering and annotation by integrating sequence and structural data, revealing conserved structures in long non-coding RNAs and targets of Roquin-1.

The paper introduces a workflow for identifying functional upstream open reading frames (uORFs) using ribosome profiling data and provides a human uORF annotation file.

The paper integrates SHAPE data into RNA-RNA interaction prediction to improve accuracy using the IntaRNA tool.

The paper identifies 39 new Type III CRISPR-Cas associated gene families using comparative genomics and presents them in an online database.

The paper presents technologies that make computational reproducibility practical by integrating package management, containerization, and workflow systems.

The paper presents a community-driven framework for interactive data analytics training in life sciences, enabling researchers to learn through web-based computation without software installation.

MechRNA predicts lncRNA mechanisms by integrating RNA-RNA and RNA-protein interaction predictions to identify potential targets and mechanisms for cancer-related lncRNAs.

The paper introduces uvCLAP, a non-radioactive method for identifying in vivo RNA-binding protein targets, showing it is reliable and comparable to other methods.

This study examines the dynamics of the cardiac myocyte epigenome during development and in chronic heart failure, revealing distinct epigenetic programs in health and disease.