News

In our paper "Network medicine-based epistasis detection in complex diseases: ready for quantum computing", we combine combinatorial optimization on graphs with quantum computing to identify SNPs involved in epistatic interactions.

In our paper "Guiding questions to avoid data leakage in biological machine learning applications", we present 7 questions that should be asked to prevent data leakage when constructing machine learning models in biological domains.

In our paper "ZEB1-mediated fibroblast polarization controls inflammation and sensitivity to immunotherapy in colorectal cancer", we investigate how the transcription factor ZEB1 promotes plasticity of cancer-associated fibroblasts that can shield tumours from the immune system. Great collaboration ...

On July 1, Nicolai Meyerhöfer started as a doctoral researcher in the BIONETS lab. He will work on graph neural network models for prediction of tissue-specific protein-protein interactions.

In our joint retreat with the DaiSyBio group (TUM) in Peschiera del Garda, we had inspiring talks and discussions on hot topics and off topics in computational systems biology, ran two hackathons, and engaged in a series of team building events.

In our paper "Efficiently Labeling and Retrieving Temporal Anomalies in Relational Databases", we show how to flag and retrieve temporal inconsistencies in healthcare information systems.

In our paper "Drugst.One — a plug-and-play solution for online systems medicine and network-based drug repurposing", we present a versatile plugin to turn any tool that outputs a list of genes or proteins into a feature rich, drug repurposing web tool with interactive network visualization.

In our paper "Inference of differential gene regulatory networks using boosted differential trees", we present BoostDiff, a tree-based method for identifying differences in transcriptional regulation between two conditions.

In our paper "Cracking the black box of deep sequence-based protein-protein interaction prediction", we show that the near-perfect performances reported for deep learning methods for PPI prediction can be entirely attributed to data leakage.

The kick-off meeting for the DyHealthNet project, held in Feldthurns (Italy) from February 1-3, 2024, was a collaborative effort involving the project partners from the University of Bozen-Bolzano, the Eurac Research Institute, the FAU Erlangen, and the TU Munich. The gathering facilitated a compreh...