jianglst / pregnancy_cfRNA

Repository to reproduce analyses from manuscript titled "Noninvasive blood tests for fetal development predict gestational age and preterm delivery"

pregnancy_cfRNA

Repository to reproduce analyses from manuscript titled "Noninvasive blood test for fetal development predict gestational age and preterm delivery"

To run the code

Install the dependencies

R packages

• plyr
• dplyr
• ggplot2
• reshape2
• cowplot
• corrplot
• caret
• randomForest
• edgeR
• limma
• locfit
• splines
• KernSmooth
• statmod
• MASS
• exactRankTests
• pROC
• effsize

Python packages

• seaborn
• numpy
• pandas
• matplotlib

Set up directory structure

The code expects the following directory structure:

• Main directory contains the following files and subdirectories:
  • Files:
    • load_data.R (Located in common_scripts folder)
    • plot_theme.R (Located in common_scripts folder)
    • counts_gene_lists.RData (Located in common_scripts folder)
  • Sub-directories titled as followed:
    • raw_data - contains 9 csv files
    • data_consolidation - contains 4 R scripts
    • fig_1 - contains 1 R script
    • fig_2 - contains 4 sub-sub-directories
    • fig_3 - contains 2 sub-sub-directories
    • supplement - contains 5 R scripts

Download raw data and relevant code

• Data can be found in the folder raw_data.
• All relevant code for each figure can be found in the folder with the corresponding name

To run code

Please note that you must change the variable mainDir in any script to reflect the main directory (See section titled "Set up directory structure")

From scratch:

• Download the raw data files and process them using the script merge_clean_data.R prior to running any other script.
• Build ML models using the script build_models.R in the folder fig_2/c_d

Using processed data:

• Download counts_gene_lists.RData located in the folder common_scripts prior to running any other script.
• Download saved model objects from the folders fig_2/c_d/models and fig_2/e/models