It is a Python tool for a quick in silico exploratory analysis of metabolic interactions in microbial communities. This code is built to study Urinary Tract Infection, but can be easily customized.
In particular, it performs a series of steps:
- build indices of complementarity and competition, based on metabolic gene annotations (KEGG orthologs).
- compare these indices with the actual experimentally-derived growth data.
- rough selection of pathways that can distinguish the best between (lack of) increase of growth.
The repository has 2 main initial folders: src and data. It also includes additional folders
that will contain the resulting tables (created) and figures (figures).
Note that the data contains a previous genome selection.
To run the pipeline, call src/main/main.py.
It calls several functions, that can be commented out if needed:
create_KOs_binary_matrix: from files derived from e.g. BLASTKoala, it creates clean binary tables for each strain.mat_to_pandas: script customized for our experimental results. Feel free to edit.create_interaction_matrices: calls a different script for each index. We included the ones in the text. Feel free to remove/add new ones as desired.big_matrix_create: creates a "big matrix" for each index, consisting of all pairs in the X-axis, and all pathways in the Y-axis. The cells contain the relevant interaction value.call_selections: calls the pathway selection methods. Again, if you feel that a different selection strategy would improve your results, it is easy to remove/add a new method.
To create the figures, call the different scripts in src/visualization/.
Before running them, make sure that you have the required tables derived from running certain
steps from the main pipeline.
The requirements for this repository are listed in requirements.txt.
They can easily be installed using e.g. pip: pip install -r requirements.txt.
The pipeline is coded in Python 3.8, whereas the figures are created in R 3.6.
The code is part of a publication that can be found here (link).