mengyin / vashr Goto Github PK

Is there a way or 'easy' modification to specify degrees of freedom for each entry of sehat separately (so that errors with more degrees of freedom get regularized less)?

Hi,

When I was trying to install your method I ran into an error which I think involves sed somehow:

install_github("mengyin/vashr",build_vignettes=TRUE)
Downloading GitHub repo mengyin/vashr@master
from URL https://api.github.com/repos/mengyin/vashr/zipball/master
Installing vashr
'/usr/local/Cellar/r/3.3.1/R.framework/Resources/bin/R' --no-site-file --no-environ
--no-save --no-restore --quiet CMD build
'/private/var/folders/9g/qzlm1ycs71sddrdzrw26hssc0000gn/T/RtmpDehXH4/devtools16d71abde138/mengyin-vashr-ee070b0'
--no-resave-data --no-manual

• checking for file ‘/private/var/folders/9g/qzlm1ycs71sddrdzrw26hssc0000gn/T/RtmpDehXH4/devtools16d71abde138/mengyin-vashr-ee070b0/DESCRIPTION’ ... OK
• preparing ‘vashr’:
• checking DESCRIPTION meta-information ... OK
• installing the package to build vignettes
• creating vignettes ... ERROR
  /usr/local/Cellar/r/3.3.1/R.framework/Versions/3.3/Resources/bin/R: line 193: /usr/local/Library/ENV/4.3/sed: No such file or directory
  ERROR: option '-e' requires a non-empty argument

Please let me know what further information I can give you to help address this issue.

Hi there,

There seems to be a versioning issue with ashr_2.1-25 for the function comppostprob. I was running the RNA-Seq part of your vignette and when I ran:

 fit.vash <- vash(sehat=sehat, df=fit$df.residual[1], 
             betahat=betahat, scale=fit$stdev.unscaled[,2])

I got the error:

 'comppostprob' could not be found.'

In the updated ashr I guess the function was renamed to comp_postprob, but it also takes in a different number of parameters such that when I tried the above after changing line 129 in vash.R with the updated function name and rebuilding I got the following error:

 Error in comp_postprob(mix.fit$g, rep(numeric(0), n), sehat[completeobs],  : 
   unused arguments (sehat[completeobs], df) 

Since comp_postprob takes in 2 variables, comp_postprob(m (mixture distriubtion with k components), data) I tried changing line 129 of vash.R to:
postpi.se[completeobs,] = t(comp_postprob(mix.fit$g, sehat[completeobs]))

and got the error

  Error in data$x : $ operator is invalid for atomic vectors .

Could you trouble-shoot this code for the updated ashr version?