Software

Welcome to Cordax, a structure-based machine learning approach that explores sequence determinants of amyloid propensity beyond their traditional limits.

Protein solubility is adapted to endogenous protein abundance in the cell where protein folding is also assisted by multiple chaperones. During recombinant protein production, purification and storage proteins are frequently handled at concentrations that are several orders of magnitude above their physiological concentration, often resulting in protein aggregation.

Tango is a computer algorithm for prediction of aggregating regions in unfolded polypeptide chains, specifically designed to predict hydrophobic beta-sheet aggregates, regardless of whether their macroscopic appearance is amorphous or amyloid-fibril like.

Waltz is a computer algorithm for prediction of amylogenic regions in protein sequences, trained from a large set of experimentally characterised amyloid forming peptides.

We have compiled a database that compiles exhaustive data on protein quality control and homeostasis in E. coli, integrating many experimental studies on chaperone interactions, protein abundance, protein folding, etc.

The SNPeffect project aims to assemble relevant biological information on - mainly coding nonsynonymous - human SNPs in one central database. Properties that can be examined in this aspect are among others differences in protein folding, aggregation, stability, interaction energy changes and changes in free energy upon mutation.

StAmP (Structural analysis of Amyloid Polymorphs) is a public repository for structural analysis of full-length amyloid fibril core structures.

WALTZ-DB 2.0 is a database for characterizing short peptides for their amyloid fiber-forming capacities. The majority of the data comes from electron microscopy, FTIR and Thioflavin-T experiments done by the Switch lab. Apart from that class of data we also provide the amyloid annotation for several other short peptides found in current scientific research papers. Structural models of the potential amyloid cores are provided for every peptide entry.

MadraX is a force field implemented as a differentiable PyTorch module and is capable of estimating the energies that stabilize protein folding. This module can be used in the same way as a Pytorch layer, which takes as input a protein structure and returns a tensor containing the Gibbs energies. The library is deep learning-ready.

The FoldX plugin for YASARA is a software package to access and run FoldX commands in YASARA, a molecular-graphics, -modeling and -simulation program that finally makes it really easy to answer your biological question

FoldX is an empirical force field for the rapid calculation of mutational free energies in proteins and nucleic acids which relies directly on structure-activity data from protein-engineering experiments to calculate interaction energies.

Limbo is a position specific prediction algorithm for identifying chaperone binding sites in proteins. It is based on a position-specific scoring matrix (PSSM) trained from in vitro peptide binding data and structural modelling.

The PyUUL tool converts biological macromolecules to data structures compatible with machine learning to help with deep learning biological structures.