Tophat2 on Windows 7.

[Coryza]

Hi,

For some (not-logical reasons) we are building our RNA pipeline on Windows. This is mainly due to biological people prefering to work with Windows. So i'm trying to install Tophat2 on windows. Tophat2 requires 2 libraries: Samtools and Boost.

- I installed Cygwin 64 with all available packages.
- Successfully compiled linux samtools to windows samtools.

I'm currently compiling boost, and whenever I try to compile Tophat2 it says it can't find Boost. So something is going wrong here. Does anyone have the instructions of how to compile boost under cygwin, and then compile Tophat2?

Thanks!

[blancha]

First of all, this is a terrible idea.
Here is an attempt at answering your question, even though the entire project should be abandoned.

After installing the Boost libraries, did you give TopHat's ./configure script the --with-boost argument specifying the location of the Boost libraries?

[Zapages]

Why can't you try dual boot Linux and Windows.

Or have Linux with emulated Windows 7. That is what we have decided on and we are biology/chemistry/biochemistry oriented lab.

Aside from this, have you thought about using Geneious for Tophat/Geneious. I know it uses valuable resources. Also please show your biologists iPlant Collaborative and Galaxy.

[Richard Finney]

What are the first few error messages?

[Coryza]

Point is, I'm only there for one more week, so I don't have any time to do something else. I've succesfully compiled Boost now, and I'm stuck with samtools.

checking for boostlib >= 1.38.0... yes
checking for bamlib... configure: error: We could not detect the bam libraries version or higher). If you have a staged bam library (still not installed) plese specify $BAM_ROOT in your environment and do not give a PATH to --with-bam otion.

Using Cygwin command in tophat2 directory: ./configure --with-bam G:/samtools-0.1.9 (containing /lib/ and /include/bam/).

[Coryza]

First of all, this is a terrible idea.
Here is an attempt at answering your question, even though the entire project should be abandoned.

After installing the Boost libraries, did you give TopHat's ./configure script the --with-boost argument specifying the location of the Boost libraries?

I'm interested though why you think the entire project should be abandoned. If it works on Windows whats wrong with it?

[blancha]

I'm not an expert on this issue, so I won't write any more posts on this thread.

If you google bamlib, you'll find a number of users reporting the same error message when trying to install TopHat or Cufflinks, and all proposing similar solutions.

I'm also wondering if you're following the more detailed installation instructions found in the tutorial of the "Getting Start guide", which should address a number of the problems that you are experiencing: http://tophat.cbcb.umd.edu/tutorial.shtml

The instructions in the manual are not detailed enough, and only provide a link to the "Getting Started guide": http://tophat.cbcb.umd.edu/manual.shtml

To answer your question, for one, you wouldn't be experiencing any of these problems . You could just install a binary version of TopHat2 in a few minutes, by just downloading the file and uncompressing it.

The best setup for a RNA-Seq pipeline is to have a Linux server administered by a bioinformatician and accessed by end-users.
I cannot think of a scenario where it would be a good idea to install the software on a Windows machine.
A Windows personal computer will generally have insufficient CPU, RAM and storage space to work efficiently.
I also find it inconvenient to have a desktop computer tied up running programs that takes hours or days to complete.
It's much more convenient to run such programs on a server.
If you have Windows servers, you should really be switching to Linux: cheaper, more reliable, and compatible with all the bioinformatics software.

I used to try and install the software on MACs, which have a Unix-like operating system, but even then, I found it was not worth the trouble.
It was just too time-consuming.
On Unix, all the bioinformatics program just work out of the box, whereas on MACs, I often needed to install extra bits, or faced some unexpected issues, as you are experiencing now when trying to compile from source.

Biologists are able to learn the Unix command line, if they are willing to make the effort. Most labs have at least one member willing to learn how to work on Unix. To make the biologists lives easier, you can install some custom pipelines for them, or one of the increasing number of off-the-shelf pipelines.
Of, if your biologists are really afraid of the command line, you should consider one of the the ever-increasing number of GUI solutions, starting with Galaxy.

[kmcarr]

I used to try and install the software on MACs, which have a Unix-like operating system, but even then, I found it was not worth the trouble.
It was just too time-consuming.
On Unix, all the bioinformatics program just work out of the box, whereas on MACs, I often needed to install extra bits, or faced some unexpected issues, as you are experiencing now when trying to compile from source.

Mac OS X is not "Unix-like". It is Unix; certified as meeting the UNIX 03 specification since version 10.5. Contrary to your experience I find Mac OS X to be an ideal desktop platform for biologists wanting to do some bioinformatics. Installing command line tools on Mac OS X has never been any more difficult (e.g. dependency requirements, etc.) than on linux boxen. Many of the popular bioinformatics tools (e.g. Bowtie, Tophat) already have binary distributions available for Mac OS X.

[blancha]

Yes, you are right.
Mac OS 10.5 and later are Unix operating systems.

There are however tiny differences with Linux that sometimes mean that you need a bit more time to perform a task when switching from Linux to MAC (e.g. programs installed by default in most Linux distributions but not in Mac OS).

More importantly, I found it very inconvenient to tie up my desktop for an analysis. I'd rather keep my desktop available and run the heavy-duty programs on a Linux server.

But, thanks for pointing out my error. I was trying to be precise, but MAC OS 10.5 and later is a Unix operating system, and not a Unix-like operating system.