RNAtracker is a computational workflow for the identification of allele-specific RNA stability (asRS) and allele-specific RNA transcription (asRT) genes.
To reproduce the gene categorization results in our paper, please follow the steps below. The first two scripts read in .txt files of initial hyperparameter values from the data/ folder that were estimated using MLE. Please refer to Supplemental Methods for additional details.
All steps only require R (scripts have been tested on R version 4.1.0). No non-standard hardware is required. The following R packages are required:
- data.table (1.14.2)
- dplyr (1.1.2)
- VGAM (1.1-10)
The package versions are listed above, but not strictly required to be the same.
After cloning the repository (git clone https://github.com/gxiaolab/RNAtracker), simply run the following scripts from the scripts/ folder. Downloading the repo should take no more than a minute.
- Determine gene ASE status at 0h
Rscript 1_preprocess_0h.R $CELL_LINE $INPUT_DIR $RESULTS_DIR
Rscript 1_preprocess_0h.R GM12878 ../data/ ../results/
- Categorize genes as asRS, asRT, mixed, or non-ASE. We require the estimated posterior probability of the assigned gene state to be ≥ 0.95 or to be above the first tercile of that state's distribution of posterior probabilities across all genes predicted to be that state.
Rscript 2_categorize_gene.R $CELL_LINE $INPUT_DIR $RESULTS_DIR
Rscript 2_categorize_gene.R GM12878 ../data/ ../results/
- For timepoints in gene categorizations that reflect ASE (e.g. 0h, 2h, and 6h for asRT; 2h and 6h for asRS), we require at least half of the SNVs for each gene to exhibit allelic imbalance in a consistent direction (e.g. the ref allele has either increased/decreased expression across both replicate 1 and replicate 2). This script will output a .txt file of the final gene categorizations and their SNVs.
Rscript 3_post_categorization_filter.R $CELL_LINE $INPUT_DIR $RESULTS_DIR
Rscript 3_post_categorization_filter.R GM12878 ../data ../results
The above steps should take no more than half an hour to run on a desktop computer.