Open zkamvar opened 9 years ago
I think microsatellite markers are not suitable for outlier loci detection.
On Tue, May 12, 2015 at 1:08 PM, Zhian N. Kamvar notifications@github.com wrote:
- Outlier detection
- SNP/environment
- Linkage disequilibrium methods
- Neutrality tests
By marker type:
- SNP
- Sequence
- Microsatellite
Please refer to the issue number when submitting pull requests.
— Reply to this email directly or view it on GitHub https://github.com/NESCent/popgenInfo/issues/52.
Margarita M. López-Uribe Postdoctoral Researcher Department of Entomology North Carolina State University https://sites.google.com/a/cornell.edu/mm-lopez-uribe/ https://sites.google.com/a/cornell.edu/margarita-m-lopez-uribe/
Perhaps not, but aren't they still suitable for Linkage disequilibrium methods?
yeap
On Tue, May 12, 2015 at 1:26 PM, Zhian N. Kamvar notifications@github.com wrote:
Perhaps not, but aren't they still suitable for Linkage disequilibrium methods?
— Reply to this email directly or view it on GitHub https://github.com/NESCent/popgenInfo/issues/52#issuecomment-101357891.
Margarita M. López-Uribe Postdoctoral Researcher Department of Entomology North Carolina State University https://sites.google.com/a/cornell.edu/mm-lopez-uribe/ https://sites.google.com/a/cornell.edu/margarita-m-lopez-uribe/
They are suitable for methods working on allele frequency (correlative method looking for association allele frequency/environment): the SNP/environment has to be replaced by : correlations allele frequency/environment.
I 'am developing the vignette SNP and signal of selection: I use packages in development and I have problem to create the pdf file from R Markdown : Knit HTML stopped because of the following error message:
"Quitting from lines 41-52 (SignalSelectionSNP.Rmd)
Error in contrib.url(repos, "source") :
trying to use CRAN without setting a mirror
Calls:
I feel that is is due to interactivity with R.
Any idea?
"Quitting from lines 41-52 (SignalSelectionSNP.Rmd) Error in contrib.url(repos, "source") : trying to use CRAN without setting a mirror Calls: ... withVisible -> eval -> eval -> install.packages -> contrib.url Execution halted"
It looks like you are using the function install.packages()
. I think you should avoid using it. This does bring up a good point on adding new packages to the docker, though. What do you think, @hlapp?
Yes, we should not install packages in any of the vignettes. Instead, they should be included in the Docker container with which we are rendering the site, and which therefore also serves as a reference for everyone else with regard to the environment under which all of the stuff described in the vignettes works.
As you can see, I use at the moment the following command to call the packages qvalues and outflank in my vignette:
install.packages("devtools") library("devtools") source("https://bioconductor.org/biocLite.R") biocLite("qvalue") install_github("whitlock/OutFLANK") library(OutFLANK)
What I need to do and which commands I need to use instaed?
I would recommend wrapping the installation part in a chunk that's not evaluated (eval = FALSE
) since these packages are already present in the docker container.
Here's how I would set it up:
```{r, eval = FALSE}
install.packages("devtools")
library("devtools")
source("https://bioconductor.org/biocLite.R")
biocLite("qvalue")
install_github("whitlock/OutFLANK")
```{r load_package}
library("OutFLANK")
```
It works...and I start to understand how it works....
:)
Sent from my iPhone
On Jan 8, 2016, at 23:56, Stéphanie Manel notifications@github.com wrote:
It works...and I start to understand how it works....
— Reply to this email directly or view it on GitHub.
I would recommend wrapping the installation part in a chunk that's not evaluated (eval = FALSE) since these packages are already present in the docker container.
I would recommend if at all possible to keep package installation out of the vignettes completely. How to install packages is covered well elsewhere and we can point to that, so as to what packages someone needs to run the code in a vignette can just be stated in code, or should be evident from the library()
commands. Installing packages changes someone's system, and running a vignette should never do that.
As for yourself, is it not possible to simply install those packages and then run the vignette assuming they are there and library()
is all you need?
I would recommend if at all possible to keep package installation out of the vignettes completely. How to install packages is covered well elsewhere and we can point to that, so as to what packages someone needs to run the code in a vignette can just be stated in code, or should be evident from the library() commands.
To that point, the package in question only exists on GitHub, so the installation is not as simple as install.packages()
. I think that wrapping it in eval = FALSE
and giving a small explanation for why the code chunk is there is the best option (as suggested in my previous comment)
By marker type:
Please refer to the issue number when submitting pull requests.