ShannonMets

Blood metabolome predicts gut microbiome α-diversity in humans

DOI https://doi.org/10.1038/s41587-019-0233-9

correspondence should be addressed to: lhood@systemsbiology.org, sgibbons@systemsbiology.org, or nathan.price@systemsbiology.org

The content is divided into two folders, one with the copies of notebooks used to generate main figures (as well as Supplementary Figure 2, which was originally a main figure) and one with code that can be run on the data to replicate our results.

Sections

1. Figure regeneration
2. Reproduce Results
3. Request data
4. Software Prerequisites
5. Docker Example

FiguresNotebooks folder:

These files are for regenerating the main figures from the paper. In order to run all these files, you will need to request data.

LASSO_RIDGE_METABOLOMICS_ANALYSIS.ipynb - code used to generate analysis presented in figures 1 and 6

FIGURE_2.ipynb - code to generate Figure 2

HEATMAPS_FIGURE3_SUPP_FIGURE2_SUPP_FIGURE6.ipynb - Code to generate Figure 3, Supplementary figures 2 & 6

Regression_analysis_Figure_4and5.ipynb - code used to generate analysis presented in figures 4 and 5

RF_Classification_Mets_and_Clinical_Labs.ipynb - code used to generate analysis in figure 6 and supplementary figure 3

Clinical_Labs_Proteomics_Regression_Analysis - code used to generate analysis in figure 6C, performance of clinical labs and proteomics in predicting Shannon diversity in the discovery and validation cohorts

Validation_compositional_microbiome-3 (1).csv microbiome - composition median abundance data for validation cohort

corr_df (1).csv - correlation coefficients between metabolites and microbiome in discovery cohort

corr_df_validation-2 (1).csv - correlation coefficients between metabolites and microbiome in validation cohort

corr_pval (1).csv - p-values of correlation between metabolites and microbiome in discovery cohort

corr_pval_validation-2 (1).csv - p-values of correlation between metabolites and microbiome in validation cohort

discovery_medians.csv - microbiome composition median abundance data for discovery cohort

ReplicationCode folder:

The ReplicationCode folder contains notebooks that will recreate the analysis described in the paper. In order to run all these files, you will need to request data.

Metabolomics_Shannon.ipynb

Code used to predict Shannon diversity, as well as PD whole tree and Chao1, from plasma metabolomics. The code generates an R2 and extracts mean beta-coefficients from each of the ten cross-validations. RUN this code before proceeding to the Classification_Analysis or OLS_Regression, since these codes rely on .CSV files with the LASSO identified metabolites.

Run order: 1

Needed input files:

• second_genome_2.csv
• data_discovery.csv

Generated output files:

• _40_coefs.csv
• top_11_mets.csv
• coeff_validation.csv

CL_P_Shannon.ipynb

Code used to predict Shannon diversity from proteomics and Clinical labs across the discovery and validation cohorts. The code also calculates an R2 for different combinations of omics data.

Run order - independent

Needed input files

• chemistries_git.csv
• chemistries_val_git.csv
• cleaned_proteomics.csv
• proteomics_validation_impute.csv
• data_discovery.csv

Generated output files

• None

Classification_Analysis.ipynb

Code used to classify individuals in the bottom quartile of Shannon diversity using 11 plasma metabolites or a panel of clinical labs.

Run order: 2 (run Metabolomics_Shannon.ipynb first)

Needed input files

• chemistries_git.csv
• second_genome_2.csv
• data_discovery.csv
• top_11_mets.csv

Generated output files

• None

OLS_Regression.ipynb

Code used to generate Ordinary Least Square Regression models between each metabolite and Shannon. The code depends on a list of the 40 mets identified, which is generated using the "Metabolomics_Shannon" code. Same code can be used to generate OLS models for chemistries and proteomics.

Run order: 2 (run Metabolomics_Shannon.ipynb first)

Needed input files:

• _40_coefs.csv
• data_discovery.csv

Generated output files

• supplementary_table_1.csv

Requesting data

Qualified researchers can access the above deidentified input files for research purposes. Requests should be sent to nathan.price@systemsbiology.org.

Software Prerequisites

After receiving the files from above, unzip the files and copy the csvs directly into the ReplicationCode folder. You should then start a Jupyter notebook server on your machine with a Python 3 kernel.

In order to run the code you will need to have installed recent versions (Python 3) of:

• seaborn
• scipy
• numpy
• sklearn
• matplotlib
• statsmodels

Docker Example

If you want to use Docker, the Jupyter scipy docker stack has all prerequisites installed.

Note: This assumes you have requested and received the data files (see Accessing data files ) and unzipped them to current folder.

Example workflow:

git clone https://github.com/PriceLab/ShannonMets.git .
cp NBT_data_files/*.csv ShannonMets/ReplicationCode/
docker run -p 8888:8888  --user $(id -u):$(id -g) --group-add users -v "$PWD":/home/jovyan/work
jupyter/scipy-notebook:1386e2046833