On Tue, Jan 6, 2015 at 12:23 AM, Peter Cock <p.j.a.cock@googlemail.com> wrote:
Yes ;)
Are you writing your own tool, or trying to install someone else's?
There are two basic options,
(1) Install Biopython into the default Python used when jobs are run. Simple, but manual - and you can only have one version. This is actually what we are still doing on our local Galaxy since we started before the second option as available.
(2) Install Biopython via the ToolShed, using the dependency mechanism. This allows multiple versions of Biopython (or any Python library) to be installed in parallel, and Galaxy does the setup to ensure the declared dependency is on the Python path when your script is run. e.g.
https://toolshed.g2.bx.psu.edu/view/peterjc/get_orfs_or_cdss
declares a dependency on:
https://toolshed.g2.bx.psu.edu/view/biopython/package_biopython_1_62
(which in turn depends on NumPy etc).
This reminds me, we are overdue to create a Galaxy ToolShed package for Biopython 1.65, I must do that...
Done, https://toolshed.g2.bx.psu.edu/view/biopython/package_biopython_1_65 And for tool developers on the TestToolShed too, https://testtoolshed.g2.bx.psu.edu/view/biopython/package_biopython_1_65 Peter