Help with running PAUDA?

[A_sapidissima]

Hi all - I am starting on a metegenomics project, and I would like to be able to run the "poor man's BLASTX", or PAUDA. I have installed the program following the instructions provided in the README file with the software, but I keep getting an error message. I do have the prerequisite software installed (Bowtie2 and Java runtime environment 1.6 or higher). I also have set the environment variables PAUDA_INDEX_DIR and PAUDA_TMP_DIR as described in the Readme. If I type "env" on the command line these variables are set as they should be.

When I run the following in the .../pauda/bin directory:

./pauda_build nr

where nr is a protein database in FASTA format in the same directory, I get the error message:

Error: PAUDA index directory "/dev/shm" already exists.
Please remove existing directory or specify another location

If I specify another directory I just keep getting the same error message. What am I doing wrong? Is there anyone who routinely runs this software that could provide some ideas as to what I have set incorrectly?

Thanks!

[GenoMax]

Do you have a machine with plenty of RAM (PAUDA recommends more than 16 GB)?

You must have set PAUDA_TMP_DIR to /dev/shm as noted in the readme for PAUDA. If you want to use an alternate directory you will have to reset the PAUDA_TMP_DIR to the new location (i.e. /tmp).

A handy background about /dev/shm and /tmp (http://superuser.com/questions/45342...ould-i-use-tmp).

[A_sapidissima]

Thanks for the reply. That does help me understand what specifying /dev/shm means. I am still learning how to use Linux, so this is great information.

The machine I am using has 20g total of RAM, so I don't think that is the problem, unless it matters what is used, free, and swap memory. The -m command says I have 14g used, and 6g free. 20g is listed as free swap memory, whatever that means.

I do have PAUDA_INDEX_DIR set as an environment variable. If I type env at the command line, I see an entry of "PAUDA_INDEX_DIR=/dev/shm".

I find the error message perplexing. To me the /dev/shm folder should already exist, otherwise how could the software use a location that hasn't already been created? I am wondering if there is some permissions setting that is set incorrectly? I get the same error message when logged in as root.

Another thing I noticed is that the Readme states that "The defaut directory for index files is called pauda-index and will be placed in the parent directory of pauda." This directory was never created when I unzipped the downloaded pauda software package. Is there some kind of makefile or something that I am missing, where I did not install the software correctly? It just says to download and unzip the file...

[GenoMax]

The machine I am using has 20g total of RAM, so I don't think that is the problem, unless it matters what is used, free, and swap memory. The -m command says I have 14g used, and 6g free. 20g is listed as free swap memory, whatever that means.

You should take a look at the output of

$ cat /proc/meminfo

since that would give you reliable info about how much physical memory you have. I am not sure what flavor of unix you are using but just going on what is reported as "free" is not always reliable since the OS will dynamically manage RAM as needed and it may be marked as "used" till it is needed for a different program.

I find the error message perplexing. To me the /dev/shm folder should already exist, otherwise how could the software use a location that hasn't already been created? I am wondering if there is some permissions setting that is set incorrectly? I get the same error message when logged in as root.

/dev/shm is referring to a ramdisk. It should be world writable.

Another thing I noticed is that the Readme states that "The defaut directory for index files is called pauda-index and will be placed in the parent directory of pauda." This directory was never created when I unzipped the downloaded pauda software package. Is there some kind of makefile or something that I am missing, where I did not install the software correctly? It just says to download and unzip the file...

I have not used PAUDA but based on the readme file I am not sure if the program is dynamically moving the index to /dev/shm. I suppose you could create a directory (anywhere) and then set the PAUDA_INDEX_DIR variable to point to that directory. It appears that you have to "pauda-build" to create the indexes. The readme does say that you would need 10-20 GB of main memory for this process (not sure if it will work with nr).