[Rd] Problem with R >3.0.0

[peter dalgaard]

On Aug 19, 2013, at 20:53 , Shelton, Samuel wrote:

> Dear R developer,
> 
> I am an R user and am currently having a problem with versions of R >3.0.0. We build larger correlation matrices of >30000 with pairwise correlations made. We have been using a line of code
> 
> simMat=bicor(datExpr1,use="p")

What is bicor()? From the WGCNA package? Perhaps the package is doing something incompatible with the long vector support in R 3.0.0. You need to report such queries to the maintainer. So far we have no evidence that the bug is in R itself, and you're not giving us anything reproducible to investigate.

> 
> To build the similarity matrix with datExpr1 being a matrix with genes as columns (>30000) and rows being samples. This code works fine on an iMac running OSX 10.8.5 and using R 2.15.2 and the Rblas library. When I try to run the same code with R version 3.0.0 and 3.0.1 it only partially builds the matrix. It will return correlations for samples 1:11262 but from 11262:30000 the matrix is full of 0's.
> 
> We would like to be able to use R 3.0.0 to allow us to build larger correlation matrices on our cluster but we can't use it at the moment due to this problem. My colleague has previously had a similar issue when using snow leopard and this was to do with a compatibility issue between snow leopard and the veclib. Has something changed in version 3.0.0 to give an incompatibility with the veclib?
> 

This seems entirely speculative. Please stick to the facts.


-- 
Peter Dalgaard, Professor,
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Email: pd.mes at cbs.dk  Priv: PDalgd at gmail.com