Disease associated processes, as well as therapeutic interventions, can be identified with perturbations within the interactome map of molecular interactions in the cell. Here, we introduce a mathematical framework for mapping out precisely how different perturbations influence each other in terms of 12 distinct interaction types. We performed a large-scale imaging screen of cells perturbed by diverse chemical compounds and identified a perturbome network of 242 perturbations and 1832 interactions. Our analysis of the chemical and biological features of the perturbing compounds reveals distinct molecular fingerprints for each interaction type. The interactome overlap between perturbations determines how they influence each other, either increasing or decreasing each others’ effects, or resulting in the emergence of entirely new phenotypes. Our framework can be applied to other key challenges, such as dissecting the combined impact of genetic variations or predicting the effect of a drug on a particular disease phenotype.
The following code can both be used to apply some or all aspects on own data as well as for the reproduction of the individual plots.
Most of the data extraction and analysis was performed using python 2.7. Analysis of the core periphery structure was conducted using the cpalgorithm module in python 3.7.0
Python packages included:
Python packages included:
Code was written and execuded on a MacBookPro
All python packages should be easily being installed via pip [https://pypi.org/project/pip/]
e.g. pip install
Typically time to install all packages should be between 30min and 1h depending on the amount of previously installed packages.
All data to run the individual scripts should be within the respective /data/
Raw fluorescent images are available under: https://idr.openmicroscopy.org/ [idr0069]
The majority of code runs within several minutes. Only exception is the calculation of interactions [9_Calculate_Interactions.ipynb] that can take from several hours up to one day depending on the underlying system. The final output of the individual analysis parts matches the output of the paper: "Mapping the perturbome network of cellular perturbations", with the only exception of some heuristic code sections.
The majority of all code is written in jupyter notebook format, including detailed information in markdown about the purpose and result of the individual sections of code.
In case anything is not clear please file an issue in this githib repository.