--read=FILENAME flag issues

[Brent Shepherd]

Nathan,

I am aligning sequences to the human genome, and because of its size, I thought I'd save the binary file created from the hash and then read it in on my subsequent code. I have the binary file saved, but every time I try to run code with the --read="..." flag, I get the following error:

gnumap: src/GenomeMem.cpp:192: unsigned int Genome::readGen(char*): Assertion `!"ERROR:  Invalid number of processors with MPI when reading genome\n"' failed.

Any ideas? Is it possible to use this feature for computing across multiple nodes (I'm using 30 nodes, 12 processors each)? Is the human genome too large to support this feature? Thanks for your help.

Brent

[Nathan Clement]

Brent,

Because we do some preprocessing when we read the genome in for the first time, there are some difficulties associated with implementing the --save and --read flags across multiple nodes. Because of this, I believe I don't allow it to match with multiple nodes. I'll look into it, though, and let you know.

With a single node, it takes a few hours to read in the entire genome, but I believe that it only takes about 30 minutes with 30 nodes.

Nathan