Re: [maker-devel] exonerate on Mac OS X 10.6

[Carson Holt]

You may be somehow skipping one of the steps.  Rather than trying to figure out which one, here is an even easier guide :-) Note this works on Mac OSX 10.6 and only 10.6, and you must be connected to the internet.

Open a terminal and type ->

sudo su

You will be asked to enter your user password (create one in ‘system preferences’ -> ‘accounts’ -> ‘change password’ if you don’t have one)

Then type ->

bash

Then cut and paste this next section directly into your terminal command line (make sure you get it all, this is basically my earlier instructions condensed into a single very long command line)->

cd /usr/local; rm -rf /sw; curl -O http://derringer.genetics.utah.edu/MWAS_devel/gcc; chmod a+x gcc; mkdir bin; mv -f gcc bin; export PATH=/usr/local/bin:$PATH; cd /usr/local; curl -O http://softlayer.dl.sourceforge.net/project/fink/fink/0.29.10/fink-0.29.10.tar.gz; tar -zxvf fink-0.29.10.tar.gz; cd fink-0.29.10; ./bootstrap /sw < /dev/null; /sw/bin/pathsetup.sh </dev/null; fink selfupdate < /dev/null; fink scanpackages < /dev/null; apt-get update < /dev/null; echo Y > empty; apt-get install libgettext3-dev=0.14.5-2 < empty; rm -f empty; fink install glib < /dev/null; fink install glib2-dev < /dev/null; cd ..; rm -rf fink-0.29.10*; curl -O http://www.ebi.ac.uk/~guy/exonerate/exonerate-2.2.0.tar.gz; tar -zxvf exonerate-2.2.0.tar.gz; cd exonerate-2.2.0; ./configure; make; make install; cd ..; rm -rf exonerate-2.2.0*;

It will run for about an hour as it compiles and installs necessary components.  When it is done type the following command to test exonerate ->

exonerate

You should not have exonerate installed.

Thanks,
Carson

On 7/7/10 12:18 PM, "Alessandro Riccombeni" <alessandro...@gmail.com> wrote:

Dear Dr. Holt,

I am a PhD student from UCD in Dublin, Ireland. I am currently trying
to install exonerate as part of the installation of Maker and I seem
to be incapable of making it work.
I tried to follow the instructions here:
http://groups.google.com/group/maker-devel/browse_thread/thread/2945997683e00fc3#

but I failed. I don't know where to put the gcc file (as if I type
"cd" only I still have several levels before the directory I get sent
to) and I am having trouble with fink. I do not have any "bin" folder
in sw/, so I don't have any /sw/bin/pathsetup.sh or such.

Could you tell me what I am doing wrong? Any suggestion?
THanks for your time.
Best regards,

Alessandro Riccombeni

[Carson Holt]

I have updated the long command line because one command appears to fail randomly, so I just have it execute twice.  Seems to fix it.

Open a terminal and type ->

sudo su

You will be asked to enter your user password (create one in ‘system preferences’ -> ‘accounts’ -> ‘change password’ if you don’t have one)

Then type ->

bash

Then cut and paste this next section directly into your terminal command line (make sure you get it all, this is basically my earlier instructions condensed into a single very long command line)->

cd /usr/local; rm -rf /sw; curl -O http://derringer.genetics.utah.edu/MWAS_devel/gcc; chmod a+x gcc; mkdir bin; mv -f gcc bin; export PATH=/usr/local/bin:$PATH; cd /usr/local; curl -O http://softlayer.dl.sourceforge.net/project/fink/fink/0.29.10/fink-0.29.10.tar.gz; tar -zxvf fink-0.29.10.tar.gz; cd fink-0.29.10; ./bootstrap /sw < /dev/null; /sw/bin/pathsetup.sh </dev/null; fink selfupdate < /dev/null; fink scanpackages < /dev/null; apt-get update < /dev/null; echo Y > empty; apt-get install libgettext3-dev=0.14.5-2 < empty; rm -f empty; fink install glib < /dev/null; fink install glib2-dev < /dev/null; fink selfupdate < /dev/null; fink scanpackages < /dev/null; apt-get update < /dev/null; echo Y > empty; apt-get install libgettext3-dev=0.14.5-2 < empty; rm -f empty; fink install glib2-dev < /dev/null; cd ..; rm -rf fink-0.29.10*; curl -O http://www.ebi.ac.uk/~guy/exonerate/exonerate-2.2.0.tar.gz; tar -zxvf exonerate-2.2.0.tar.gz; cd exonerate-2.2.0; ./configure; make; make install; cd ..; rm -rf exonerate-2.2.0*;

[Carson Holt]

I’m glad you were able to get exonerate installed which is very difficult on a Mac.  If you look at the code for the gcc module I included with the earlier thread, this is infact what it does internally.  It always uses -m32.  This solves some problem with many fink modules (apart from just installing exonerate) because some fink modules will ignore the GCC environmental variable when installing.

Repeatmasker runs slow with cross_match (I haven’t yet tried it with the new repeatblast they recently make available as an alternative to wublast).  RepeatMasker runs about 10-20x faster with wublast.   Try installing the newest version of repeatmasker with repeatblast as the alignment engine, maybe that will help.

The empty blastdb can be caused by a couple of issues.  First are you using NCBI blast or NCBI blast+?  Second one of your fastas might not be formatted correctly.  A very common error is a fasta files contaminated with meta returns rather than unicode endlines.  If you use the terminal editing program emacs these will show up as ‘^M’ characters in the fasta file.  Also you can send me the fasta files and I can take a look at them.

Thanks,
Carson

On 7/9/10 11:41 AM, "Alessandro Riccombeni" <alessandro...@gmail.com> wrote:

Dear Dr. Holt,

a colleague of mine, while I was waiting for your reply, suggested me
to change the configure file. I changed

GCC

to

GCC -m32

and exonerate seemed to install fine.
I ended up installing all the fundamental parts, keeping the ncbi
blast 2.2.22 I already had, as my colleagues only had a compiled
wublast for Linux.
Maker seems to work now, but I have two questions:
-RepeatMasker seems to take a lot of time: is it supposed to be the
longest step, or could I be doing something wrong? I can also see that
only 100MB of memory are being used;
-I successfully ran maker with the example data, but when I try to use
my own data I fail, getting a failed chunk at level 10, when doing
blastx, because of an empty database.
I am using 3 files:
genome sequence from species A, a multifasta with nt sequence;
transcript sequence from species B;
protein sequences from species B.

Compared to the options I used for the example data, I am not using
est evidence (Est) but only another species' transcript (altest) and I
didn't specify anything in the predictor model, where I used
est2genome for the example data. Is one of these 2 differences the
cause of an empty blast db? I don't understand the cause of the error.
Still, everything worked for the example data you provided... please
let me know if you have any suggestion, and thanks for your time (and
for maker, as well!).
Thanks,

ALessandro R.