wgmao / plier Goto Github PK

Hi! I saw that there are 18 gene sets in /data, but ListPrior() shows only 8 of them. Do you not recommend to use the other 10 prior gene sets? Also, can I get the source/reference of those gene sets? It seems like not all for them has the source information. Thanks!

Hello there, we are interested to use our own pathway file. Could you describe how you assembled e.g. canonicalPathways? We have a gmt file listing the genes in pathways from GOBP right now.
Thanks!

Hi,

I'm trying PLIER with the tutorial dataset but getting the following error:

here is the traceback:

and sessionInfo:

Thanks,
Ariel

Hi, I've been trying to run PLIER with the rseed argument and I get the following error message:

Error in Z %*% B : non-conformable arguments

Traceback shows this is occurring on line 492 of PLIER/Allfuncs.R. You can reproduce this fairly easily using the data included with the PLIER library as follows:

library(PLIER)
data(canonicalPathways)
data(dataWholeBlood)
PLIER(dataWholeBlood, canonicalPathways, rseed=1)

This occurs for any value of rseed, as long as a value is passed to the PLIER function. When no value for rseed is passed it runs fine.

I'm using R version 3.6.0 - happy to provide versions of other packages if it would be useful. Thanks!

Hi I tries to apply PLIER to a single cell datasets.

Here is the command I used,
I can provide additional information if needed!

any help appreciated!

plierREsults=PLIER(as.matrix(subset(
+                           subset(merged, seurat_clusters %in% c("0", "1", "2", "3", "4", "11")),
+                           downsample=500)@[email protected]),
+                   pathways_matrix)
Selecting common genes: 12549
Removing 151 pathways with too few genes
Computing SVD
Using rsvd
Error in qr.default(Y, complete = FALSE) : 
  NA/NaN/Inf in foreign function call (arg 1)

I want to define the principle components for defining the range of k, but I got this error:

k <- PLIER::num.pc(data=gtex_expression_matrix_cm)
Error in if ((class(data) != "list") & (class(data) != "rsvd")) { : 
  the condition has length > 1

The gtex_expression_matrix_cm is the same object used i PLIER::PLIER and works perfectly.

Hello, I couldn't find a description in the documentation but could you describe the items in the output object? I think what I'd like to extract is something like plierResult$summary, but for all of the pathways that were submitted.

Hi, I'm trying to install PLIER in an Anaconda environment using the conda skeleton cran command. Unfortunately, doing so fails unless there is at least one release for the repo. Would it be possible to add a release to make the repo compatible with conda?

Hello-

I'm running into an issue when I run through the example on your vignette (and my own data set):

When I run PLIER as per the vignette (adding in this step that is previously computed):

"plierResult=PLIER(vacDataN[cm.genes,], allPaths[cm.genes,], k=34, trace=T)"

I get the following error:

"Error in qvalue(pval) : could not find function "qvalue"
Q-value error, defaulting to BHThere are 16 LVs with AUC>0.70"

and then subsequentlly running "plotU(plierResult, auc.cutoff = 0.75, pval.cutoff = 0.01, top = 3)" gives the error:

Error in [.data.frame(plierRes$summary, plierRes$summary[, 5] < fdr.cutoff, : argument "fdr.cutoff" is missing, with no default

So then, I thought maybe I just need to include "library(qvalue)", and when I do that, I avoid the error after running

"plierResult=PLIER(vacDataN[cm.genes,], allPaths[cm.genes,], k=34, trace=T)"

but when I run the next step, "plotU(plierResult, auc.cutoff = 0.75, pval.cutoff = 0.01, top = 3)"

I still get the error:

"Error in [.data.frame(plierRes$summary, plierRes$summary[, 5] < fdr.cutoff, : argument "fdr.cutoff" is missing, with no default"

I hope this is clear. I'd love to try this package on my data. Just in case you need it, my sessionInfo is pasted below. Thanks a lot!

sessionInfo()
R version 3.4.3 (2017-11-30)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Sierra 10.12.6

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] stats graphics grDevices utils datasets methods base

other attached packages:
[1] PLIER_0.99.0 rsvd_0.9 knitr_1.19 glmnet_2.0-13 foreach_1.4.4 Matrix_1.2-12 pheatmap_1.0.8
[8] gplots_3.0.1 RColorBrewer_1.1-2

loaded via a namespace (and not attached):
[1] Rcpp_0.12.15 codetools_0.2-15 lattice_0.20-35 gtools_3.5.0 bitops_1.0-6 grid_3.4.3 plyr_1.8.4
[8] gtable_0.2.0 scales_0.5.0 KernSmooth_2.23-15 gdata_2.18.0 iterators_1.0.9 tools_3.4.3 munsell_0.4.3
[15] compiler_3.4.3 colorspace_1.3-2 caTools_1.17.1

Hi,
Please suggest if I should do a log transformation of TPM (or RPKM) values or not to run the plier. Which works better out of these two?

Hi,
Good work.
In the file man/simpleDecomp.Rd there is a closing curly bracket missing for the arguments. This creates a problem for installing the package.

Regards,

I have a quick question about this package: Can the data of mRNA of human tissue be used for this method to analysis? Thank you!

Hi @wgmao !
When I tried to ran PLIER on a small dataset (670 samples and 327 genes), I got such error:

plierResult <- PLIER(expr_input,canonicalPathways)
Selecting common genes: 124
Removing 519 pathways with too few genes
Computing SVD
Using rsvd
Done
k is set to 34
[1] 11.45349
[1] "L2 is set to 11.4534942166023"
[1] "L1 is set to 5.72674710830116"
errorY (SVD based:best possible) = 0.8997
Error in elnet(x, is.sparse, ix, jx, y, weights, offset, type.gaussian,  : 
  y is constant; gaussian glmnet fails at standardization step

When I ran PLIER on a larger dataset , PLIER can work correctly. Does these errors occur due to the small size of my dataset? Thank you!!!

Hi,
Very nice tool, but I have question to better interpret the results that I am getting.

Lets say, I run PLIER on a cohort of 100 samples, and observe that a specific pathway P1 is significantly aligned along the latent vector L1. On the other hand, we have pathway P2 which does not align with any of the latent vector (i.e AUC is very small). Then my question is, can I make any statement about the activity of pathway P1 and P2 in my 100 samples. In other words, is it correct to say that pathway P1 was active across my samples but pathway P2 was not active. Please let me know what is the right interpretation. I know that one can score P1 and P2 across samples using scores from B matrix (please correct if I am wrong), but can we bluntly say that if plier does not align a specific pathway along any of the latent vector, does it mean that pathway is inactive?

Second question is, how stable are the results if I add/remove new pathways to the prior information. If pathway P1 was significant on one of the latent vector when using N number of gene lists as prior information, will it be significant if we extend or shrink the prior information to N+n or N-n gene lists.

I would appreciate your response for these two questions

Hi!

I'd like to ask you several questions about PLIER.

1. When I computed the LVs, PLIER returned There are 26 LVs with AUC>0.70. However, I didn't find the AUC of each LV in the plierResults. Can you tell me how to calculate AUC of each LV from the results?
2. In your tutorial, when you visualized the top genes and their pathway associations, you used
   indexToPlot <- which(apply(plierResult$Uauc * (plierResult$Up < 0.001), 2, max) > 0.75) to select index of accociated pathways. Can you tell me the exact meaning of this equation?
3. Do you have pathway data using Ensembl ID? I found that all the pathways in your R package using HGNC Gene Symbol. In the website (http://gobie.csb.pitt.edu/PLIER/), it seems that we can submit gene expression matrix using Ensembl ID, but I still haven't received the mail one day after I submitted the job.

HI! I'm wondering whether you have a recommendation on the size of prior. For example, below are different prior combinations I've tried. I run PLIER on recount2 datasets (from MultiPLIER paper, https://doi.org/10.1016/j.cels.2019.04.003) containing 6,750 genes x 37,032 samples.

1. bloodCellMarkersIRISDMAP, svmMarkers, canonicalPathways (6,847 genes x 628 pathways) --> 690 out of 987 LVs are annotated
2. c2.cgp, c2.cp, c6, c7 from MSigDB v.7.0 (22,116 genes x 9,748 pathways) --> 135 out of 195 LVs are annotated
3. c6 from MSigDB v.7.0 (11,250 genes x 189 pathways) --> 430 out of 599 LVs are annotated

Thanks!