[BiO BB] Anybody knows the usage stats for different apps in sequence alignment & molecular dynamics?

[Chris Dwan (CCGB)]

> I am starving for CPU!

Would it be sufficient to provide a set of interfaces (web services,
RMI's, whatever is hip these days) to fairly standard analysis tools, or
do you need to run custom executables?

Using existing technology, it's tricky but very achievable to share our
spare cycles for known analyses with collaborators.  So long as we're able
to speak the language of "BLASTN this query vs NCBI dbEST no older than
January 1, 2004", we can provide a moderate amount of CPU to a lot of
people at minimal effort.

Would that be useful?  More concretely, who else would be willing to share
their cycles, and who would be willing to make use of such a service?

The harder problem ("~/a.out -i /data/freakishly_huge.dat >
   10.1.1.1:/export/foo.out") involves solving quite a few problems:  data
movement, binary incompatibility, privacy, and trust to name just a few.
Of these, trust is by far the largest.  Users need to trust that their
results are accurate, and providers need to trust that users are well
behaved.

Or perhaps I'm just very confused about the real problems out there in the
world.

-Chris Dwan
 The University of Minnesota