data- each subdirectory contains EC, transcript and exon counts from the following data sets:data/drosophila- simulated drosophila data from Soneson et al.[1]data/hsapiens- simulated human data from Soneson et al.[1]data/bottomly- real mouse data from Bottomly et al.[2]
R- helper functions to perform analyses and generate paper figuresref- transcript ID lookup tables and results from the Soneson et al. paper (used for comparison)
First download the simulation truth data from http://imlspenticton.uzh.ch/robinson_lab/splicing_comparison/ and place all the text files under ref/soneson_results/. Next, download the RunInfo metadata table for the Bottomly data set from NCBI. Put this file under ref/.
Download the feature counts for the Love et al. data from 10.5281/zenodo.2644723. Un-tar the data into the notebook directory.
The ec-dtu-paper.Rmd markdown notebook can now be run using knitr in RStudio. This will run the paper analyses and generate the figures under the figures folder. This analysis will be time consuming to run from start to finish, and a multi-core and high-memory machine is therefore recommended.
Please install the following R packages:
install.packages('ggplot2')
install.packages('gridExtra')
install.packages('VennDiagram')
install.packages('RColorBrewer')
install.packages('data.table')
install.packages('readr')
install.packages('dplyr')
install.packages('devtools')
source('http://bioconductor.org/biocLite.R')
biocLite('edgeR')
biocLite('DEXSeq')
biocLite('DRIMSeq')
biocLite('tximport')
devtools::install_github('mikelove/rnaseqDTU')
Under the data folder, a matrix of equivalence class results is included per data set. This was generated using the included python script create_salmon_ec_count_matrix.py. This script is designed to be run on Salmon[3] output, and can be run as follows:
python create_salmon_ec_count_matrix.py <eq_classes> <samples> <outfile>
For example:
python create_salmon_ec_count_matrix.py \
sample1/aux_info/eq_classes.txt sample2/aux_info/eq_classes.txt \
sample1,sample2 ec_matrix.txt
NOTE: for later versions of salmon, do not use the --validateMappings flag, unless you also use the --hardFilter flag. Otherwise, ECs cannot be matched correctly between samples.
A script for transforming Kallisto[4] output is also included. Kallisto must be run using the --batch parameter (see https://pachterlab.github.io/kallisto/manual), with all samples concurrently. The usage for the script is as follows:
usage: create_kallisto_ec_count_matrix.py [-h]
ec_file counts_file samples_file
tx_ids_file out_file
positional arguments:
ec_file Kallisto equivalence class file (matrix.ec).
counts_file Kallisto counts file (matrix.tsv).
samples_file Kallisto samples file (matrix.cells).
tx_ids_file File containing one transcript ID per line, in same order as
the fasta reference used for kallisto.
out_file Output file.
numpy and pandas is required to run both scripts.
[1] Bottomly, D., Walter, N. A. R., Hunter, J. E., Darakjian, P., Kawane, S., Buck, K. J., … Hitzemann, R. (2011). Evaluating gene expression in C57BL/6J and DBA/2J mouse striatum using RNA-Seq and microarrays. PLoS ONE, 6(3). https://doi.org/10.1371/journal.pone.0017820
[2] Soneson, C., Matthes, K. L., Nowicka, M., Law, C. W., & Robinson, M. D. (2016). Isoform prefiltering improves performance of count-based methods for analysis of differential transcript usage. Genome Biology, 17(1), 1–15. https://doi.org/10.1186/s13059-015-0862-3
[3] Patro, R., Duggal, G., Love, M. I., Irizarry, R. A., & Kingsford, C. (2017). Salmon provides fast and bias-aware quantification of transcript expression. Nature Methods, 14(4), 021592. https://doi.org/10.1038/nmeth.4197
[4] Bray, N. L., Pimentel, H., Melsted, P., & Pachter, L. (2016). Near-optimal probabilistic RNA-seq quantification. Nature Biotechnology, 34(5), 525–527. https://doi.org/10.1038/nbt.3519