This repository contains the code written for the bioinformatic study in A pathogen-encoded signaling receptor mediating host-like interactions through intrinsic disorder (temporary title) (2021).

We use sequence-based predictors to assess and compare the amount of structural disorder in proteins from intestinal pathogens. Per-amino acid scores were calculated, aggregated per-protein and used to classify them.

Effector collections from Enteropathogenic and Enterohemorrhagic Escherichia coli (EPEC and EHEC, respectively) and Citrobacter rodentium, the standard small animal A/E model, were studied and compared to reference proteomes from the corresponding species. Additionally, the human reference proteome was analyzed to provide a perspective of the eukaryotic host.

The predictions were done with DISOPRED 3 and IUPred 1.0. Since DISOPRED is more computationally expensive the calculations were run in a cluster, so the code here only runs the IUPred predictions.

Most functions are imported from the src/aepathbactdis.py module. Self-descriptive Python scripts are run on each fasta folder for each predictor (DISOPRED and IUpred 1.0 in short and long modes).

As a preparatory step, map_taxa.py builds json files mapping the corresponding protein IDs (Uniprot format) to the corresponding taxon numbers and ultimately reference proteomes.

The processing pipeline then has two stages:

• Prediction of structural disorder of protein sequence collections (run_iupred.py).
• Aggregation and classification of the individual proteins (aggregate_*.py).

Jupyter notebooks are used for analysis and plotting (see /notebooks). The resulting summary files are saved to the respective aggregated data folders, and plots are saved to a /figures folder (see Data availability).

The associated data (sequence collections, prediction scores, aggregated values and classifications) is accessible here.

Assuming you have Python installed, the required libraries can be installed via pip install -r requirements.txt.

IUPred 1.0 is needed to run the predictions. It can be downloaded here and installed where desired. The path is optionally provided to run_iupred.py as an argument.

This software is licensed under the MIT license.

• Miguel Arbesú ([email protected])
• Tiago Cordeiro ([email protected])
• Andreas Zanzoni ([email protected])