The generation module

With the generation module, you can generate a conformational ensemble for your protein of interest.

📝 Setting up your parameters file

Before generating an ensemble, you must create a parameters file either:

Below you can find a description of the minimum required parameters.

Parameter Description
Job Name Name for generated files and folders.
Input Structure/Sequence Full-length structure
available or Full IDP
Path to structure/sequence in .pdb/.txt format. A UniProt accession
code can also be provided, in which case the corresponding structure
from the AlphaFold Protein Structure Database is retrieved.
Folded domains
available, missing IDRs
(Uses modelling module)
Path to FASTA file(s) containing the sequences of all
(folded + disordered) protein domains, from N- to C-terminal.
Can be provided either in FASTA or Multi-FASTA format.
Path to PDB file(s) containing the structures of all folded
protein domains, from N- to C-terminal. Can be provided
as a single PDB file with multiple MODEL entries.
Size of Ensemble Desired number of conformers in the generated ensemble.
Database(s) Mapping of database IDs to the path of their respective
database files. Currently supported file formats
include .pkl, .csv and .parquet.
Sampling Target(s) Protein regions to be targeted for conformational sampling.
You must assign for each desired sampling target a protein
chain, a range of residue numbers, a database ID to sample
from (matching one defined in Databases(s)) and a sampling
mode ('Tripeptide', if neighbouring residue information
is to be considered, or 'Single Residue' otherwise).
Output Path Path to desired output directory. A directory named Job Name
will be created here, with all generated files and folders.

Generate a conformational ensemble

To generate an ensemble, simply provide Ensemblify with the path to your parameters file.

(ensemblify_env) $ ensemblify generation -p parameters_file.yml

Get suggestions for target sampling regions

In case you are unsure which regions of your protein you want to sample, you can provide Ensemblify with your protein’s UniProt accession.

This will trigger automatic queries of useful structure/disorder related databases (MobiDB, TED, AFDB), outputting informative reports about the protein’s domain annotations.

For example, if you were interested in modelling full-length Galectin-3 (UniProt accession P17931):

(ensemblify_env) $ ensemblify generation -st P17931