Integrated omics reveals actionable drivers of bioactive variation in US milk

This repository contains data, analysis code, processed outputs, and figure assets for the DMI milk omics study.

Database access

The database is available online at dairymolecules.com. An account for browsing the data is available upon request.

Study overview

Directory overview

• Code/: analysis scripts organized by figure or supplemental figure groups.
• Data/: raw and reference datasets used by the analyses.
• ProcessedData/: intermediate and final processed tables generated by scripts.
• Figures/: exported figure files used in the manuscript.

Software versions

• R: tested with R 4.2.2
• Python: tested with Python 3.11

R package dependencies

The following R packages are imported by scripts under /home/runner/work/DMI/DMI/Code:

Biostrings, ComplexHeatmap, MOFA2, PubChemR, RColorBrewer, RCurl, XML, basicdrm, car, circlize, colorspace, data.table, doParallel, dplyr, emmeans, fastcluster, foreach, ggplot2, ggraph, ggrastr, ggsankey, gsubfn, jsonlite, lavaan, lme4, mapdata, maps, nnls, patchwork, plotrix, scales, stringr, this.path, tidygraph, usmap, xml2

Install from CRAN:

install.packages(c(
  "PubChemR", "RColorBrewer", "RCurl", "XML",
  "basicdrm", "car", "circlize", "colorspace", "data.table", "doParallel",
  "dplyr", "emmeans", "fastcluster", "foreach", "ggplot2", "ggraph",
  "ggrastr", "ggsankey", "gsubfn", "jsonlite", "lavaan", "lme4", "mapdata",
  "maps", "nnls", "patchwork", "plotrix", "scales", "stringr", "this.path",
  "tidygraph", "usmap", "xml2"
))

Install Bioconductor packages:

if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")
BiocManager::install(c("Biostrings", "MOFA2", "ComplexHeatmap"))

Python package dependencies

The following third-party Python packages are imported by scripts under /home/runner/work/DMI/DMI/Code:

matplotlib, numpy, pandas, scipy, scikit-learn, tqdm

Only the following Code workflows currently contain Python scripts:

• /home/runner/work/DMI/DMI/Code/SuppFig15_CompoundPCA
• /home/runner/work/DMI/DMI/Code/Fig5_SuppFig12_29_SNP_miRNA_CompoundCorr/Python_CorrelationEvaluation

Install with pip:

pip install matplotlib numpy pandas scipy scikit-learn tqdm

Regenerating figures and supplementary figures

To regenerate figures/supplementary figures, run the analysis scripts inside each subdirectory under /home/runner/work/DMI/DMI/Code.

• Each subdirectory corresponds to one figure or supplemental figure workflow.
• Some workflows are R-based and some include Python scripts.
• Outputs are written to ProcessedData/ and Figures/ according to each workflow.

Workflow execution dependencies

Some workflow directories depend on outputs generated by other workflow directories, so they should be run in dependency order rather than independently.

Examples:

• Code/1_ConcentrationTablePreparation.R should be run first to generate shared processed matrices used by multiple workflows.
• Code/SuppFig10_11_JerseyBreedAssociation/CowbreedCompositionRatioEstimation.R must run before BreedRatioBoxplot.R and Heatmap_SNPs.R.
• Code/Fig3_SuppFig6_17_Bioactivity/ generates processed bioactivity outputs used by workflows such as Code/Fig4_SuppFig8_24_RegionAnalysis/, Code/Fig5_SuppFig12_29_SNP_miRNA_CompoundCorr/, and Code/SuppFig30_MediationAnalysis/.
• Code/SuppFig18To23_MOFA/RunMoFA.R produces SNP matrices required by Code/SuppFig25To28_SNPConstructCompoundProfiles/.

For detailed input/output dependencies and run-order notes, please check the README.md inside each Code subdirectory.

Notices

1. The full execution datasets are not included in this repository; data can be made available upon request.
2. Additional manual manipulation is required for some files and will be specified in the input/output section of the corresponding subdirectory README.md.