Skip to main content

Machine-learning prediction of residues driving homotypic transmembrane interactions.

Project description


The Transmembrane HOmodimer Interface Prediction Algorithm (THOIPA) is a machine learning method for the analysis of protein-protein-interactions.

THOIPA predicts TM homodimer interface residues from evolutionary sequence information alone.

THOIPA was designed to complement experimental approaches, and also energy-based modelling of TM homodimers.

What does thoipapy do?

  • download protein homologues with BLAST

  • extract residue properties (e.g. residue conservation and polarity)

  • trains a machine learning classifier

  • validates the prediction performance

  • creates heatmaps of residue properties and THOIPA prediction


pip install thoipapy

THOIPA standalone can be currently installed on Linux.

A version of THOIPA in a docker container is in development.

See Wiki for full details.


We recommend the Anaconda python distribution, which contains all the required python modules (numpy, scipy, pandas,biopython and matplotlib). THOIPApy is currently tested for python 3.6.

Pip should automatically install the pytoxr package of Mark Teese.

THOIPApy depends on the command-line programs phobius and freecontact. Both of these are only available for Linux. THOIPApy itself has been tested on several different systems running Windows and Linux.

Development status

The code has been extensively updated and annotated for public release. However is released “as is” with some known issues, limitations and legacy code. The THOIPA standalone predictor is currently available to use. The settings file and databases used for THOIPA training are not yet released.

Usage as a standalone predictor

For TMD interface residue predictions of a protein of interest, we recommend running THOIPA as a standalone program via Docker, as described in the Wiki .

THOIPA can also be installed in Linux and used as a standalone predictor: * The operating system needs to have freecontact, phobius, and NCBI_BLAST installed. * The biopython wrapper for NCBIblast should be installed.

import thoipapy
protein_name = "ERBB3"
predictions_folder = "/path/to/your/output/folder"
blastp_executable = "blastp"
phobius_executable = "phobius"
freecontact_executable = "freecontact"
thoipapy.run_THOIPA_prediction(protein_name, TMD_seq, full_seq, predictions_folder, phobius_executable, freecontact_executable)

Example Output

Create your own machine learning predictor

Details on how to train THOIPA on your own datasets will be released after publication.

import THOIPApy
settings = r"D:\data\THOIPApy_settings.xlsx"


THOIPApy is free software distributed under the permissive MIT License.


THOIPApy is not yet officially published. However, feedback regarding the installation and usage of the standalone version is appreciated. Simply email us directly, or initiate an issue in Github.


For contact details, see the relevant TU-Munich websites:

Author: Bo Zeng of the Frishman lab, TU-Munich, Weihenstephan Campus

Further coding and supervision: Mark Teese of the Langosch lab, TU-Munich, Weihenstephan Campus


Citation to be added. Full Credits: Bo Zeng, Yao Xiao, Dmitrij Frishman, Dieter Langosch, Mark Teese

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

thoipapy-0.0.2.tar.gz (114.5 kB view hashes)

Uploaded source

Built Distribution

thoipapy-0.0.2-py3-none-any.whl (126.0 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page