Björn Wallner

A generalisable framework to inject distance information into Alphafold-like structure predictors

This paper introduces a method to inject distance information into AlphaFold-like structure predictors to improve structure prediction accuracy.

AFsample3: Generating and selecting multipleconformational states with Alphafold3

AFsample3 improves protein conformational sampling by generating and selecting multiple states, outperforming AlphaFold3 and AFsample2 in accuracy and diversity.

DockQ v2: Improved automatic quality measure for protein multimers, nucleic acids, and small molecules

The paper introduces DockQ v2, an improved tool for assessing the quality of modeled biomolecules, including protein interfaces, nucleic acids, and small molecules.

AFsample2: Predicting multiple conformations and ensembles with AlphaFold2

AFsample2 improves prediction of protein alternative states and conformational ensembles by reducing co-evolutionary signals, enhancing structural diversity and model quality.

Structural characterization of the Pseudomonas Aeruginosa MexR-mexR repressor-operator complex: a small-angle X-ray and neutron scattering perspective

This study characterizes the structure of the MexR-mexR repressor-operator complex using SAXS and SANS, revealing DNA bending and flexibility between binding sites.

Towards understanding of allostery in MALT1: a possible role of interdomain motions as revealed by NMR and AlphaFold

The paper investigates the dynamic behavior of MALT1 using NMR and AlphaFold to reveal interdomain motions involved in allosteric regulation.

Unmasking AlphaFold: integration of experiments and predictions in multimeric complexes

This study introduces AF_unmasked, a method that integrates experimental data with AlphaFold predictions to improve the accuracy of large protein complex structures.

AFsample: Improving Multimer Prediction with AlphaFold using Massive Sampling

AFsample improves multimer prediction by using AlphaFold with stochastic perturbation and massive sampling, achieving higher DockQ scores in CASP15.

dgram2dmap: Extraction, visualisation and formatting of distance constraints from AlphaFold distograms

dgram2dmap converts AlphaFold distograms into distance maps for visualization and downstream analysis.

Improving Peptide-Protein Docking with AlphaFold-Multimer using Forced Sampling

This study benchmarks AlphaFold-Multimer for peptide-protein interaction prediction and shows that forced sampling improves docking accuracy.

InterPepScore: A Deep Learning Score for Improving the FlexPepDock Refinement Protocol

The paper introduces a graph neural network called InterPepScore to enhance the Rosetta FlexPepDock refinement protocol for peptide-protein docking, improving model quality on a benchmark.

InterPepRank: Assessment of Docked Peptide Conformations by a Deep Graph Network

InterPepRank ranks peptide-protein docking decoys using a deep graph network, improving the selection of high-quality models.

Critical Assessment of Protein Intrinsic Disorder Prediction

The paper evaluates 43 methods for predicting intrinsically disordered regions and binding regions in proteins, finding that deep learning methods outperform older physicochemical approaches but binding prediction remains challenging.

InterLig: a fast and accurate software for ligand-based virtual screening

InterLig is a software for ligand-based virtual screening that compares small molecules using 3D atom alignments, outperforming existing methods on a benchmark.

InterPep2: Global Peptide-Protein Docking with Structural Templates

InterPep2 predicts peptide-protein interaction structures using structural templates, outperforming existing methods on a benchmark dataset.

rawMSA: End-to-end Deep Learning Makes Protein Sequence Profiles and Feature Extraction Obsolete

The paper introduces rawMSA, a deep learning approach that eliminates the need for sequence profiles and features by directly processing raw multiple sequence alignments to predict protein structural features.

Predicting protein-peptide interaction sites using distant protein complexes as structural templates

This paper introduces InterPep, a computational pipeline that predicts protein-peptide interaction sites using structural templates and machine learning, achieving 50.7% rank 1 accuracy and 69.3% top five accuracy.

Topology independent structural matching discovers novel templates for protein interfaces

The paper introduces InterComp, a topology and sequence-order independent method for structural comparisons of protein interfaces, showing it can identify similar interfaces in 283 of 586 targets.

Improved protein model quality assessments by changing the target function

The paper improves protein model quality assessment by retraining ProQ3D with different target functions, finding contact-based methods better for identifying best models and S-score better for global quality correlation.

Methods For Estimation Of Model Accuracy In CASP12

The paper describes methods for estimating model accuracy in protein structure predictions, showing improvements in single model accuracy over previous methods.

Proteus: A Random Forest Classifier to Predict Disorder-to-Order Transitioning Binding Regions in Intrinsically Disordered Proteins.

Proteus predicts disorder-to-order transitioning residues in intrinsically disordered proteins using a random forest classifier, outperforming existing methods with higher true positives and precision.

InterPred: A pipeline to identify and model protein-protein interactions

InterPred predicts and models protein-protein interactions from sequence using structural modeling and a novel random forest classifier, outperforming existing methods.

The positive inside rule is stronger when followed by a transmembrane helix.

The study shows that the positive inside rule enhances translocon recognition of marginally hydrophobic transmembrane helices when followed by a transmembrane helix, but is not sufficient alone.