Code for generating figures in "Protein structure-based gene expression signatures"

Rahman, R et al. 2021

Instructions

To generate figures from the article you can run the scripts found in each directory during an interactive R session, where each line of the script is sent to the R terminal.

Each script for each figure is standalone and can be run individually. However all scripts require the data directory in the figure directory to be populated.

There are a few utility scripts included in this repository to parse GTEx data or to create and train autoencoders in R.

This code has been validated to run on the following platforms: Ubuntu 20.04 LTS, Ubuntu 18.04 LTS and Windows Subsystem for linux. Your mileage may vary for other platforms.

Data directory

All of the code references a figures/data/ directory This is a directory that houses data used to generate figures.

Due to the size of the generated data (>4GB), it is not included in this repo, but is available to download from here: http://iyengarlab.org/dtoxs/PNAS-sGES.tar.bz2

Please extract the files and place folder into figures/data/. The directory structure should be as follows:

data/
├── TISSUES-harmonizome
│   ├── inte-gl
│   ├── results
│   └── rna-gl
├── archs
│   ├── ARCHS4_r
│   ├── embeddings
│   └── structural-signatures
│       └── formatted
├── dtox
├── gtex
│   ├── stable_signature_data
│   │   ├── selected_gene_lists
│   │   └── stable_signature_data_old
│   └── structural-signatures
├── humanprotatlas
│   ├── genelists
│   ├── raw
│   └── results
├── l1000
├── msigdb
│   ├── genelists
│   ├── raw-gmt
│   └── results
│       └── ss_output
├── reconstruction_errors
└── rocs

Directories

figures/ Contains R code and scripts used to generate figures for the paper.

figures/figure-1/ code for figures 1C (supplemental figure S2)

figures/figure-3/ code for figures 3A-D (supplemental figure S4)

figures/figure-4/ code for figure 4 (supplemental figure S6)

figures/figure-5/ code for figures 5A-D (supplemental figures S8-11)

figures/figure-6/ code for figures 6A-E (supplemental figures S12-17)

figures/figure-7/ code for figures 7A-C

scripts/autoencoder scripts to train autoencoder models and output reconstruction errors for test data from GES or sGES data

utilities/ simple shell scripts to run strucutural signatures on GTEX and ARCHS4 data

figures/data generated data that is analyzed by the scripts

Dependancies

R libraries:

tools

BBmisc

RColorBrewer

Rtsne

data.table

dplyr

ggdendro

ggplot2

gplots

grid

gridExtra

iheatmapr

keras

magrittr

pROC

plotly

plyr

ranger

rhdf5

stringr

tidyr

tidyverse

vegan

Python Libraries (will be installed during R keras installation )

Tensorflow

Keras

Dependancy installation

Use the following code to install the R dependancies. Python tensorflow will also be installed during the installation of the repsective R versions

library("tools","BBmisc","RColorBrewer","Rtsne","data.table","dplyr","ggdendro","ggplot2","gplots","grid","gridExtra","keras","magrittr","pROC","plotly","plyr","ranger",,"stringr","tidyr","tidyverse","vegan")

if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")

BiocManager::install("rhdf5", "iheatmapr")

Contact

For any issues relating to this repository contact the corresponding authors: rayees(dot)rahman(at)icahn(dot)mssm(dot)edu or avner(dot)schlessinger(at)mssm(dot)edu