PRECISE is an RNA-seq compendium for Escherichia coli from the Systems Biology Research group at UC San Diego.
The following data files are available in the data folder:
- log_tpm_full.csv: Expression levels for all genes in E. coli
- log_tpm.csv: Expression levels for 3,923 genes in E. coli (noisy genes have been removed)
- log_tpm_norm.csv:
log_tpm.csvcentered to reference condition (WT on glucose M9 media) - metadata.csv: Experimental metadata (e.g. strain descriptions, carbon source etc.) for all 278 conditions in PRECISE
- gene_info.csv: Descriptive characteristics of genes, including location, operon, and COG group
- TRN.csv: Known regulator-gene interactions from RegulonDB 10.0
- S.csv: Gene coefficients for each iModulon
- A.csv: Condition-specific activities for each iModulon
- curated_enrichments.csv: Detailed information on iModulons and their linked regulator(s)
- imodulon_gene_names.txt: List of gene names in each iModulon
- imodulon_gene_bnumbers.txt: List of genes (as b-numbers) in each iModulon
A conda environment for this code has been provided here
To generate robust independent components for a dataset, execute the run_ica.sh script:
run_ica <filename.csv>
where <filename.csv> is a comma-separated file of gene expression. Data must be centered using a reference condition (See data/log_tpm_norm.csv for an example)
Additional options are included as flags. Decreasing tolerance (e.g. -t 1e-3) will reduce runtime, but will also reduce the final number of independent components.
The Jupyter notebook exploratory_analysis.ipynb walks users through the data files and includes a few small functions for interrogating iModulons.
Python 3.6 or greater
Conda environment specifications are listed in environment.yml
Versions of scikit-learn above 0.20.3 cause an error when performing ICA.