A computational toolbox associated with the manuscript entitled GEMCAT — A new algorithm for gene expression-based prediction of metabolic alterations. Cite using: https://www.biorxiv.org/content/10.1101/2024.01.15.575710v1
Note: We are still refining the tool. Particularly, GEMCAT does not yet provide guidance for significance of predicted changes or any other measure of prediction quality. We suggest filtering the predictions for consistency. We do not recommend pre-filtering the transcriptomics and proteomics data based on significance as this is affecting the network coverage which might negatively impact the prediction quality as genes/proteins not present in the dataset should be unchanged.
We tested the package for compatibility with Python >= 3.10 on Ubuntu and Windows.
Install from pip:
pip install gemcat
Or clone the repository and install GEMCAT from there using:
pip install .
Use a single file containing per-gene fold-changes to calculate the resulting differential centralities:
gemcat <./expression_file.csv> <./model_file.xml> -e <column_name> -o <result_file.csv>
Make sure the .csv file is either comma- or tab-delimited.
column_name is the name of the column in the file containing the fold-change.
Alternatively, use two files (or one file) with expression values for condition and baseline:
gemcat <./condition_file.csv> <./model_file.xml> -e <condition_column_name> -b <./baseline_file> -c <baseline_column_name> -o <result_file.csv>
If you do not have a model file ready, some models can be automatically accessed using their names:
gemcat ./expression_file.csv <model_name> -e column_name -o <result_file.csv>
Model names currently supported are:
Currently, GEMCAT supports models in SBML, JSON, and MAT formats.
Important points to remember: Your gene or protein identifiers should be the first column of the expression file. Make sure the gene or protein identifiers in your expression data file exactly match those in the model. A results list of all 1.0 is a sure sign of no identifier matching.
Positional arguments:
- expression file path
- model file path
All parameters:
-e --expressioncolumn name of column containing condition expression data
-b BASELINE, --baseline file containing baseline expression data
-c BASELINECOLUMN, --baselinecolumn name of column containing baseline expression data
-v VERBOSE, --verbose enables verbose output
-o OUTFILE, --outfile write output to this file
-l LOGFILE, --logfile write logs to this file
import gemcat as gc
results = gc.workflows.workflow_standard(
cobra_model: cobra.Model,
mapped_genes_baseline: pd.Series,
mapped_genes_comparison: pd.Series,
adjacency = gc.adjacency_transformation.ATPureAdjacency,
ranking = gc.ranking.PagerankNX,
gene_fill = 1.0
)
This will return the changes in centrality relative to the baseline in a Pandas Series. When using fold-changes as the mapped expression, use a vector of all ones as a comparison.
GEMCAT is modular, and its central components can easily be swapped out or appended by other components adhering to the specifications laid out in the module base classes (primarily adjacency transformation, expression integration, and ranking components). All classes inheriting from the abstract base classes laid out in the modules are exchangeable.
The core of the package is the GEMCAT model structure that contains the model data, integrates the workflow, and calculates the results.
Different approaches can be used to calculate adjacency in the networks. We offer alternatives and a platform to create custom algorithms for the model.
Module covering the mapping of gene values onto reactions in the model via gene product rules. Providing different algorithms along with a platform to create alternatives.
Module providing ranking algorithms for the models along with a platform to include custom algorithms.
The workflow module contains example workflows. To customize the workflow to your needs simply copy the provided functions and switch out the desired steps.
Command-line interface for GEMCAT.
Input and output functions that create GEMCAT models from different sources.
Contains common utility functions used throughout the package.
Functions to verify data integrity.
Functionality for automatic downloading, storing, and retrieving of common models.
You can run all local tests with pytest .. Default behavior is to also run integration tests, which takes time.
You can exclude slow running tests by using pytest . -m "not slow".
These slow running tests are integration tests with real world data and will take 10-30s each according to your hardware.
To run tests, make sure you have git lfs installed and all the Tests are running.
Make sure to run isort and black to have properly formatted code.
The CI pipeline in GitHub will check with isort, black, and pytest.