Molecular subtyping of hypertensive disorders of pregnancy

Overview

This repository contains code to reproduce the statistics and figures for "Molecular subtyping of hypertensive disorders of pregnancy".

Code Files

Each code files generates results as described: 1. Stats.Rmd: R markdown for generating statistics in manuscript 2. Figure_2abc.R: R script for creating Figures 2a-2c 3. Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb: Jupyter notebook for creating Figures 2d, 2e, 3, 4a, 4b, 4d, and Extended Data Figures 1 & 3, Supplementary Table 1, and analysis for days spent in the NICU 4. Figure_4c.R: R script for creating Figure 4c 5. Figure_5a.ipynb: Jupyter notebook for creating Figure 5a 6. Figure_5b.R: R script for creating Figure 5b 7. Figure_Extended_Data_2ab_4_5.R: R script for Extended Data Figure 2a and 2b, 4, and 5 8. SI_proximity_to_delivery.R: R script for "Evaluation of Collection Proximity to Delivery versus HDP Severity" in supplemental information.

Dependencies

Dependencies are listed per code file.

Stats.Rmd

• R 4.2.2
• tidyverse 2.0.0
• knitr 1.45
• arrow 10.0.1
• jsonlist 1.8.7
• correlation 0.8.4
• psychometric 2.4
• effectsize 0.8.8

Figure_2abc.R, Figure_4c.R, Figure_5b.R

• Requirements listed for Stats.Rmd
• patchwork 1.2.0
• roxygen2 7.3.1
• colorspace 2.0-3
• ggtext 0.1.2

Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb

• python 3.12.4
• numpy 1.26.4
• pandas 2.2.0
• matplotlib_venn 1.1.1
• scipy 1.14.0
• matplotlib 3.8.4
• seaborn 0.13.2
• sklearn 1.5.1
• statsmodels 0.14.2
• Figures2d2e34a4b4dExtendedData13SupplTbl1NICU_helper.py (dependencies outlined above)

Figure_5a.ipynb

• python 3.10.13
• jupyter 1.0.0
• numpy 1.26.4
• pandas 1.5.3
• matplotlib 3.8.3
• seaborn 0.13.2
• sklearn 1.4.1.post1
• fig5a_helper.py (dependencies outlined above)

Figure_Extended_Data_2ab_4_5.R

• R 4.4.1
• cogena 1.21.2
• fgsea 1.30.0
• ggplot2 3.5.1
• msigdbr 7.5.1
• tidyverse 2.0.0
• ggpubr 0.6.0

SI_proximity_to_delivery.R

• R 4.2.2
• tidyverse 2.0.0
• arrow 10.0.1
• Hmisc 5.1-3

Installation Guide

Installing dependencies for R and R markdown

Figure_2abc.R, Figure_4c.R, Figure_5b.R, SI_proximity_to_delivery.R, and Stats.Rmd

To install dependencies, perform the following: 1. Install R 4.2.2 from https://cran.r-project.org/bin/windows/base/old/ 2. Install RStudio from https://posit.co/download/rstudio-desktop/ 3. Install `devtools` via install.packages("devtools"). 4. Install the listed dependencies via calls to devtools::install_version("PACKAGE", version = "VERSION", repos = "http://cran.us.r-project.org"), where PACKAGE is the package name and VERSION is the package version

Figure_Extended_Data_2ab_4_5.R

To install packages, run ```

CRAN packages

listofcranpackages <- c("fgsea","ggplot2", "msigdbr", "tidyverse", "ggpubr") new.packages <- setdiff(listofcranpackages, installed.packages()[,"Package"]) if(length(new.packages)) install.packages(new.packages)

Devtools packages

devtools::install_github("zhilongjia/cogena")

Bioconductor packages

if (!require("BiocManager", quietly = TRUE)) install.packages("BiocManager") BiocManager::install("Biobase", force = TRUE)

attach libraries if necessary/desired

read_library <- function(...) {

obj <- eval(substitute(alist(...)))

#print(obj)

return(invisible(lapply(obj, (x) library(toString(x), character.only=TRUE))))

}

read_library(cogena, fgseq, ggplot2, msigdbr, tidyverse)

```

Installing dependencies for Jupyter notebooks

To run the Jupyter notebook, ensure minimal dependencies are installed: * python (see versions above) * conda (see versions above) * jupyter (see versions above)

Then install the library requirements listed above with conda install.

Inputs

The code is setup to run on the manuscript data. Manuscript data are available with a signed data use agreement to protect identifiable data. Please contact research@mirvie.com.

To run the code on your own data, the file structure must match as described below and column labels should be edited as appropriate in the scripts.

Input files required by most scripts:

Most scripts (excluding Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb and Figure_5a.ipynb) require sample_data.feather, which is a feather formatted sample-wise dataframe containing samples as rows and gene names and metadata information as columns. The gene expression values here are corrected log2cpm.

Input file for Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb is sample_data_scaled.feather, which is a feather formatted sample-wise dataframe containing samples as rows and gene names and metadata information as columns. The gene expression values here are scaled and corrected log2cpm.

Input files for Figure_5a.ipynb are samples from clinical validation that are not accessible to the modeling team. Code can be run with alternative input files for testing.

Input files additionally required by Stats.Rmd

Stats.Rmd additionally requires genes_space.json, a JSON containing gene names comprising the search space.

Input files additionally required by Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb

In addition to the Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU_helper.py script, this notebook requires 3 additional input data files: * genes_space.json: a JSON containing gene names comprising the search space.

The following 3 input files are feather formatted gene-wise dataframes containing genes as rows and p-values, effect sizes, and analysis labels as columns to generate the following figures: * fig3_de_data.feather: input for generating Figure 3 * fig4ab_de_data.feather: input for generating Figures 4a, 4b * suppl_tbl1_data.feather: input for generating Supplementary Table 1

Input files additionally required by Figure_Extended_Data_1abc.R

This script additionally requires a GMT annotation file.

Running the Code

Note that input files must be in the same directory as the script/notebook that is run.

Running R and R markdown

Expected run times should complete in less than a minute.

Instructions for running Stats.Rmd

1. Install needed dependencies as listed above. 2. Open the file in Rstudio. 3. Hit the "knit" button in Rstudio.

Instructions for running Figure_2abc.R, Figure_4c.R, Figure_5b.R, and SI_proximity_to_delivery.R

1. Install needed dependencies as listed above. 2. Open the file in Rstudio. 3. Execute code in order. Can execute the entire file from the R console via source("SCRIPTNAME.R", echo = TRUE).

Instructions for running Figure_extended_data_2AB_4_5.R

The script can be sourced from R console or in Rstudio environment.

Running Jupyter Notebooks

Instructions for running Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb or Figure_5a.ipynb:

1. Start the Jupyter server in the folder containing the notebook. 2. From the Jupyter browser, click on the *.ipynb to open it. 3. In the notebook, change Kernel to the conda environment created with the required dependenices installed. 4. Click on "Kernel" > "Restart & Run All"

Expected Run Times * Figures_2d_2e_3_4a_4b_4d_Extended_Data_1_3_Suppl_Tbl_1_NICU.ipynb: Expected run times should complete in less than a minute. * Figure_5a.ipynb: Expected run time is less than 1 hour.

Software License

Creative Commons license CC BY-NC v4.0