search

hartleys / JunctionSeq

The JunctionSeq R package is a powerful tool for testing and visualizing differential usage of exons and splice junctions in next-generation, high-throughput RNA-Seq experiments.
28 stars 15 forks source link

DESeq2 version collision #36

Open gfucile opened 6 years ago

gfucile commented 6 years ago

Hello,

I'm using the latest version of JunctionSeq (1.5.4), and have generated the following error using DESeq2 versions 1.20.0 (R 3.5.0) and 1.18.1 (R 3.4.2):

fitDispersionFunction() Starting
...
NOTE: Autodetected DESeq2 v1.15+. Adding simple weights to C++ function calls.
Error in .Call("DESeq2_fitDisp", PACKAGE = "DESeq2", ySEXP, xSEXP, mu_hatSEXP,  : 
  "DESeq2_fitDisp" not available for .Call() for package "DESeq2"
Calls: runJunctionSeqAnalyses ... adapted.estimateDispersionsMAP -> fitDispWrapper -> fitDisp -> .Call
Execution halted

I suppose the interim solution is to downgrade to DESeq2 pre v1.15 ?

Hopefully it's not too complicated to update JunctionSeq to play nicely with current and future DESeq2 versions..?

Many thanks for your help and this excellent tool.

sum732 commented 6 years ago

I am facing the similar issue: fdf(): 'Shrinking' fitted and feature-specific dispersion estimates. NOTE: Autodetected DESeq2 v1.15+. Adding simple weights to C++ function calls. Error in .Call("DESeq2_fitDisp", PACKAGE = "DESeq2", ySEXP, xSEXP, mu_hatSEXP, : "DESeq2_fitDisp" not available for .Call() for package "DESeq2"

There version for DESeq2 I have is 1.20.0, JunctionSeq:1.5.4 with R: 3.5.0

hartleys commented 6 years ago

The github version sometimes conflicts depending on which version of R/DESeq2 you are running.

I recommend using the Bioconductor version that goes with your current R version. In other words, install using the command:

 source("https://bioconductor.org/biocLite.R")
 biocLite("JunctionSeq")
gfucile commented 6 years ago

With which version(s) of DESeq2 is the latest version of JunctionSeq compatible?

hartleys commented 6 years ago

The GitHub version is compatible with Bioconductor 3.4, and DESeq2 v1.10.1. It might be compatible with 3.5, I'm not sure. It definitely is incompatible with any of the versions found in bioconductor 3.6 and higher.

I recommend using the Bioconductor version. I haven't updated the GitHub version to match because it includes C++ code that requires compilation. Bioconductor pre-compiles all those source files on linux, windows, and OSX. Without this pre-compilation, installing packages that include C++ code from source is non trivial and requires the several external tools which may not be installed on users' machines. On linux it's usually fine but on OSX and windows it's a pain.

I keep the bioconductor version up-to-date. Use that one.

pratarora commented 3 years ago

any solution to this? I am using DeSeq2 v1.32.0 and JunctionSeq v1.5.4. Bioconductor doesn't support JunctionSeq now. Is it possible to run this anyway?

s223102 commented 1 year ago

Is there another way to use JunctionSeq other than downgrading? As pratarora wrote, it seems like Bioconductor doesn't support JunctionSeq now.

I run

ramikheireddine commented 11 months ago

Error in .Call("DESeq2_fitDisp", PACKAGE = "DESeq2", ySEXP, xSEXP, mu_hatSEXP, : "DESeq2_fitDisp" not available for .Call() for package "DESeq2" I am having the same issue. I am using R: 4.3.1 DeSeq v1.42.0 JunctionSeq v1.5.4 Bionconductor 3.18

Any solution for this