Machine learning detection of unstable antibiotic heteroresistance in E. coli

The pipelines are created using Snakemake

Data analysis and modelling are performed using R and tidyverse.

Snakemake pipelines

Main pipeline

File: workflow/snakefile.smk

Purpose: assembling and annotating E. coli genomes (resistance genes, IS elements, direct repeats) from both short and long sequencing reads.

Configuration file: workflow/config.yaml

To run the pipeline short and long reads should be in resources/data_raw/{strain}/short/ and resources/data_raw/{strain}/long/ directories.

DAG:

Phylogeny pipeline

File: workflow/phylogeny.smk

Purpose: phylogenetic analysis of the samples including 27 reference strains.

Configuration file: workflow/config_phylogeny.yaml

Analysis of mutants

File: mutants.smk

Purpose: analysis of the HR mutants.

Configuration file: workflow/config_mutants.yaml

DAG:

Data analysis and machine learning

To get the same versions of packages, use renv::restore()

For feature generation see notebooks/modelling/features.qmd.

For exploratory data analysis of the features, see file notebooks/modelling/EDA.qmd,

For training and validation procedures, see notebooks/modelling/training_and_validation.Rmd,

For analysis of the models, see notebooks/modelling/models_analysis.Rmd

Features table: notebooks/modelling/data/features_strain.csv

Raw reads availability

The raw sequencing reads used in this project will be available from NCBI's SRA under BioProject PRJNA1165464