Descrease annotation running time by spliting contigs into separated files (maker)

Question

[Rox]

Hi everyone !

I am using the tool maker to annotate my genome. The problem is that this tool is not using several cores to do the task.

So I was wondering, what about splitting my genome assembly contigs into several five (let's said 32, my number of CPU). For example, if I have 64 contigs, the 2 first one goes on one file, the 2 second ones goes on second file, and it goes on. If I launch maker, with the same amount of evidences for each file, will that be suitable ? Or does maker kind of use all the file at once to create gene models ?

The first step of maker, on my whole assembly, last something like 12 days (I have a lot of proteins evidences). I want to speed up that time by spliting the file into several part, where maker will work on in the same time. Does that seem acceptable ? Anyone already tried that ?

Thanks !

Roxane

[shenwei356]

I found some things about running MAKER with multiple threads . http://weatherby.genetics.utah.edu/MAKER/wiki/index.php/MAKER_Tutorial

 -cpus|c <integer> Tells how many cpus to use for BLAST analysis.
             Note: this is for BLAST and not for MPI!

If you want to install the MPI version of MAKER, you need to have Perl
  compiled for threads and install threads 1.67 or higher.

  You will also need to install the following perl modules from CPAN.
   *DBI
   *DBD::SQLite
   *Proc::ProcessTable
   *threads (Required by MPI and accessory scripts)

Follow the instructions to set this file up, and start the mpi environment with mpdboot. 
Then run maker through the MPI manager mpiexec.

    mpdboot
    mpiexec -n 2 maker </dev/null

mpiexec is a wrapper that handles the MPI environment. 
The -n 2 flag tells mpiexec to use 2 cpus/nodes when running maker.
For a large cluster, this could be set to something like 100. 
You should now know how to start a MAKER job via MPI.