Last updated: 2019-01-12

workflowr checks: (Click a bullet for more information)

  • ✔ R Markdown file: up-to-date

    Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

  • ✔ Environment: empty

    Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

  • ✔ Seed: set.seed(20180714)

    The command set.seed(20180714) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

  • ✔ Session information: recorded

    Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

  • ✔ Repository version: 7951c3c

    Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

    Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated: 
    
Ignored files:
    Ignored:    .DS_Store
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    docs/.DS_Store
    Ignored:    docs/figure/.DS_Store

Untracked files:
    Untracked:  analysis/gd_notes.Rmd

Unstaged changes:
    Modified:   analysis/brain.Rmd
    Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
Expand here to see past versions:
    File Version Author Date Message
    Rmd 7951c3c Jason Willwerscheid 2019-01-12 workflowr::wflow_publish(“analysis/flashier_bench.Rmd”)

GTEx eQTL data

First, I fit the “strong” subset of SNP-gene association statistics used in Urbut, Wang, Carbonetto, and Stephens (2018) (the strong.z dataset found here). I fit five FLASH factors with normal-mixture priors using flashr and using flashier with backfit.order set to "dropout", "sequential", and "montaigne".

The data is a dense 16k x 44 matrix that takes up 7.0 MB of memory when loaded into R. I used the broadwl partition of the midway2 RCC cluster with 4 CPUs and 8 GB of memory, and I used Gao Wang’s monitor_memory.py script to test memory usage, as recommended in Peter Carbonetto’s large-scale data analysis tutorial.

GTEx eQTL data
VMS (GB) RSS (GB) Greedy (s/iter) Backfit (s/iter) Backfit iter Obj diff Time (min)
flashr 0.67 0.42 0.56 0.41 195 1894 2.22
dropout 0.48 0.30 0.41 0.41 182 1894 1.98
sequential 0.40 210 1894 2.15
montaigne 0.41 324 1894 2.97

Montoro et al. 3’ scRNA-seq data

Next, I fit the droplet-based 3’ scRNA-seq dataset analyzed in Montoro et al. (2018) (the data can be obtained here). I performed a log-plus-one transform of the data, then I fit five FLASH factors using normal-mixture priors.

The data matrix is 18k x 7k and takes up 1011 MB of memory when loaded into R as a dense matrix. However, only 9.3% of entries are nonzero, so the data can also be loaded as a sparse Matrix object, in which case the data takes up 143 MB of memory. I fit FLASH objects to the larger matrix object using the same four approaches used to fit the GTEx dataset, then I fit a FLASH object to the sparse Matrix object using flashier with backfit.order = "dropout" (flashr does not support objects of class Matrix). All fits were performed on the broadwl partition of the midway2 RCC cluster using 4 CPUs and 32 GB of memory.

Montoro 3’ scRNA data
VMS (GB) RSS (GB) Greedy (s/iter) Backfit (s/iter) Backfit iter Obj diff Time (min)
flashr 18.4 18.1 23.49 15.64 250 27002 101.97
dropout 3.9 3.7 1.91 1.92 285 26982 11.27
sequential 2.71 250 27109 13.45
montaigne 1.82 100 13004 5.20
sparse 1.1 0.9 0.90 0.88 285 26982 5.22

Montoro et al. full-length scRNA-seq data

Finally, I fit the larger full-length scRNA “PulseSeq” dataset from Montoro et al. (2018). The dataset is about ten times larger than the droplet-based scRNA-seq dataset, so it was not feasible to use flashr. I again fit five factors using normal-mixture priors, and I set backfit.order = "dropout".

The dataset is 22k x 66k, with 9.3% of entries not equal to zero, and occupies 1.49 GB of memory when loaded into R as a sparse Matrix object. (I did not attempt to fit a larger matrix object.) The fit was again performed on broadwl using 4 CPUs and 32 GB of memory.

Montoro full-length scRNA data
VMS (GB) RSS (GB) Greedy (s/iter) Backfit (s/iter) Backfit iter Obj diff Time (min)
sparse 6 5.8 7.74 3.9 334 300023 27.23

Session information

sessionInfo()
R version 3.4.3 (2017-11-30)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS High Sierra 10.13.6

Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/3.4/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.4/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
 [1] workflowr_1.0.1   Rcpp_1.0.0        digest_0.6.18    
 [4] rprojroot_1.3-2   R.methodsS3_1.7.1 backports_1.1.2  
 [7] magrittr_1.5      git2r_0.21.0      evaluate_0.12    
[10] highr_0.7         stringi_1.2.4     whisker_0.3-2    
[13] R.oo_1.21.0       R.utils_2.6.0     rmarkdown_1.10   
[16] tools_3.4.3       stringr_1.3.1     xfun_0.4         
[19] yaml_2.2.0        compiler_3.4.3    htmltools_0.3.6  
[22] knitr_1.20.22

This reproducible R Markdown analysis was created with workflowr 1.0.1