plot_transcript_heatmap doesn't return plot, returns gtable object

[mchimenti]

Hello:

I'm attempting to plot a transcript heatmap with:

p<- plot_transcript_heatmap(so, head(so_tab_wt_geno, n = 10)$target_id, 'tpm', cluster_transcripts = FALSE)

However, nothing is plotted and a "large gtable" object called 'p' is created.

p TableGrob (1 x 1) "arrange": 1 grobs z cells name grob 1 1 (1-1,1-1) arrange gtable[layout]

Calls to "print(p)" do not work, neither does 'gridExtra::grid.arrange(p$gtable)'.

I can see from the plotting code on github that the function modifies the gtable and then is supposed to return the plot: https://github.com/pachterlab/sleuth/blob/master/R/plots.R } invisible(dev.off())

modify the column labels with the x_axis_angle

subtracting from 360 degrees to get it to align well without modifying

anything else

p$gtable$grobs[[3]]$rot <- 360 - x_axis_angle

this sends the graphic back

gridExtra::grid.arrange(p$gtable) }

But in my case, I do not get a plot back. Any help is appreciated!

---

sessionInfo() R version 3.4.3 (2017-11-30) Platform: x86_64-apple-darwin15.6.0 (64-bit) Running under: OS X El Capitan 10.11.6

Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.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] parallel stats4 stats graphics grDevices utils datasets methods base

other attached packages: [1] bindrcpp_0.2.2 dplyr_0.7.6 org.Hs.eg.db_3.5.0 AnnotationDbi_1.40.0 IRanges_2.12.0
[6] S4Vectors_0.16.0 Biobase_2.38.0 BiocGenerics_0.24.0 biomaRt_2.34.2 cowplot_0.9.2
[11] ggplot2_3.0.0 sleuth_0.30.0

loaded via a namespace (and not attached): [1] progress_1.1.2 tidyselect_0.2.4 purrr_0.2.5 reshape2_1.4.3 rhdf5_2.22.0 colorspace_1.3-2
[7] yaml_2.1.18 blob_1.1.1 XML_3.98-1.11 rlang_0.2.1 pillar_1.2.2 glue_1.2.0
[13] withr_2.1.2 DBI_1.0.0 bit64_0.9-7 RColorBrewer_1.1-2 matrixStats_0.53.1 bindr_0.1.1
[19] plyr_1.8.4 stringr_1.3.0 zlibbioc_1.24.0 munsell_0.4.3 gtable_0.2.0 devtools_1.13.5
[25] memoise_1.1.0 curl_3.2 Rcpp_0.12.17 scales_0.5.0 bit_1.1-12 gridExtra_2.3
[31] digest_0.6.15 stringi_1.2.3 grid_3.4.3 tools_3.4.3 bitops_1.0-6 magrittr_1.5
[37] lazyeval_0.2.1 RCurl_1.95-4.10 tibble_1.4.2 RSQLite_2.1.0 tidyr_0.8.1 pkgconfig_2.0.1
[43] pheatmap_1.0.10 data.table_1.11.4 prettyunits_1.0.2 assertthat_0.2.0 httr_1.3.1 R6_2.2.2
[49] compiler_3.4.3

[warrenmcg]

Hi @mchimenti,

This is a mistake in the documentation (stating the result is a ggplot object, which it was in the previous iteration of this method; instead it is indeed a gtable object). It is intended to be printed directly to a device like this:

## Example: print plot to a PDF file
pdf('figure.pdf')
plot_transcript_heatmap(so, head(so_tab_wt_geno, n = 10)$target_id,
                        'tpm', cluster_transcripts = FALSE)
dev.off()

## Or if you want it to print to the console directly
plot_transcript_heatmap(so, head(so_tab_wt_geno, n = 10)$target_id,
                        'tpm', cluster_transcripts = FALSE)

## if you have stored the result as 'p',
## then using gridExtra::grid.arrange should print the plot to a device
p <- plot_transcript_heatmap(so, head(so_tab_wt_geno, n = 10)$target_id,
                             'tpm', cluster_transcripts = FALSE)
pdf('figure.pdf')
gridExtra::grid.arrange(p)
dev.off()

If this does not work, let me know. If it does work, we can update the documentation to correct the mistake and make it clear how to use the method to users. If you would prefer a ggplot2 object, Is there a use case where a gtable object would not work?

[warrenmcg]

Hi @mchimenti,

A further question: if you downgrade your version of R to before 3.4.3, do you still experience the same issue? Another user found this to be the case. If so, then this is a deeper issue with grid and gridExtra with later versions of R.

[emb356]

Another error I believe is related to this change: When I try to print both plot_transcript_heatmap() and plot_sample_heatmap() to a pdf (like in your first example), only the first plot will show up in the pdf. I did not have this issue with the previous version of Sleuth.