novalign MPI

[jgSoton]

hi,

I'm using novoalignMPI and found this:

Release 2.05.17 - 5th November 2009
----------------------------------
novoalign
1. When running multithreaded default output is now asynchronous with read files. If you need output in sync with reads add option -oSync to the command line.

Should I be using this option? I'm not sure what it means.

Thanks,
Jane

[zee]

Hi Jane

This option is only valid for standalone novoalign and not the MPI version. So all this means is that if you want to get the read order the same in the alignment file as in the reads file then you use this option. For the MPI version this is not possible because the message passing and return states are asynchronous and it would require an algorithm change.

[jgSoton]

Thanks for the reply.

Do I understand correctly that there is a multithreaded stand alone version of novoalign which this command should be used for. And a separate MPI version which the -o sync command is not applicable?

I was confused thinking that "multithreaded novalign" and "novoalign MPI" were the same thing.

Thanks,
Jane

[zee]

That is correct. The multithreaded version is limited to a single server.

[dawe]

BTW, asynchronous data writing should speed up things. The only issue may be that running novoalign twice on the same data will give you two results ordered in different ways. If you are going to check data integrity using some hashing (such as md5), you may have some problems.

d