Ulrike Stege

The paper introduces Mut2Dens, a model that improves protein secondary structure prediction by incorporating missense mutation data, reducing prediction inaccuracies in mutated structures.

This study benchmarks network inference algorithms for deriving metabolic relationships from lipidomic and metabolomic data, revealing challenges in interpretation despite some ability to distinguish metabolic states.

The paper evaluates how missense mutations affect protein structure prediction accuracy, finding that current methods overestimate structural changes and struggle with stable mutations and complex proteins.

The paper predicts energetically favorable secondary structures of coronavirus frameshift elements using hierarchical folding and experimental data, identifying stable motifs and transitions relevant to -1 PRF regulation.

CParty computes a hierarchically constrained partition function for pseudoknotted RNA structures in cubic time, enabling stochastic analysis of pseudoknotted RNA ensembles and revealing kinetic features of SARS-CoV-2 targets.

The paper introduces a benchmark framework for evaluating deep learning models in promoter recognition and finds that current models lack sufficient precision on eukaryotic genomic sequences.

The paper improves secondary structure prediction of SARS-CoV-2 -1 PRF pseudoknot and identifies structural similarities with SARS-CoV and MERS-CoV pseudoknots.