Code repository for: Transcriptomic profiling and machine learning uncover gene signatures of psoriasis endotypes and disease severity
Rider AC, Grantham HJ, Smith GR, Watson DS et al. on behalf of the PSORT consortium
The raw and adjusted gene count data from our RNA-seq analysis, along with associated clinical data are available at Array Express under accession number E-MTAB-14509.
The RNA-Seq data may be visualised and further explored through a R Shiny web interface.
Username: psort
Password: tower squid ramp
In order to reproduce the analyses in this paper you will need to download the RNA-Seq data and clinical data from ArrayExpress (see link above). You will also need the supplementary data from the paper. Predicted cell type fractions from single cell deconvolution analysis are also available in this repository in the following directory: Cell_Type_Correlations/paper_data
This repository contains several R scripts and workbooks which cover the core analyses in this paper. These are oraganised into several directories and below we provide details about which scripts and workbooks correspond to which figures in the paper. Rows in the table below marked as preliminary analyses should be run before those for generating figures.
| Figure | Link to script or workbook | Details |
|---|---|---|
| Preliminary analysis | link | Runs WGCNA to identify gene modules |
| Preliminary analysis | link | Runs ICA to identify latent factors |
| Preliminary analysis | link | Runs PCA to show effect of Discovery/Replication cohort and its mitigation |
| 1A, 1B | link | Calculates correlations between modules/factors and traits |
| 1A, 1B | link | Creates imput files for Metascape; functional annotations from Metascape were used to create descriptive module and factor names |
| 1A, 1B | link | Creates descriptive module and factor names which are used as annotations in the module/factor-trait correlation heatmaps |
| 1A, 1B | link | Creates module/factor-trait correlation heatmaps |
| 1C | link | Plots exemplar module/factor-trait correlations |
| 2A, 2B | link | Carries out BMI differential expression analysis and creates BMI and PASI association heatmap |
| 2C | link | Creates exemplar gene-level BMI and PASI association plots |
| 3 | link | Carries out BMI endotype analysis |
| 4 | link | Provides overview of machine learning methodology |
| 5, 6 | link | Carries out PASI differential expression analysis |
| 5, 6 | link | Creates PASI volcano plots |
| 5, 6 | link | Creates Metascape heatmaps |
| Supplementary figure | Link to script or workbook | Details |
|---|---|---|
| 4 | link | Runs module preservation analysis between skin and blood |
| 4 | link | Creates plots to visualise the module preservation analysis results |
| 6 | link | Creates cell type correlation heatmaps |
| 7 | link | Plots expression of single cell marker genes from Hughes et al. (2020) |
| 10 | link | Creates cell type correlation heatmaps |
| 18 | link | Carries out PASI differential expression analysis |
| 18 | link | Creates PASI volcano plots |
| 18 | link | Creates Metascape heatmaps |