I don't really like the idea of handling it at display time either, but there
are all sorts of legacy data sets w/ empty cells, which certainly came from
spreadsheets of clinical data w/ empty cells (not sure why there are a variety
of ways that's getting parsed by R and thus a variety of ways it is being
inserted into the DB though - I figured there'd be only *one* 'empty'.

There could presumably be some sort of filter when loading the datasets to
convert empty cells in the pData to "NA".

ERROR reading file: msg111
Possibly a access right configuration problem.
[Errno 13] Permission denied: '/extra/trackers/sgdi/db/files/msg/0/msg111'

Will soon be checking in code that converts the various "empty" values that I
found (None, "" and " ") to "NA" when displaying.

I would agree on it being a display issue.  It would be pretty easy to convert
any blanks to 'NA'.  While I doubt he'd disagree, I'll run this one past SR as
well when I'm pestering him about the other thing - ie which the biologist types
would prefer to see.

ERROR reading file: msg105
Possibly a access right configuration problem.
[Errno 13] Permission denied: '/extra/trackers/sgdi/db/files/msg/0/msg105'

Weird - in a dataset I have locally there are some blank values (not necesarily
bad data input, some samples don't have all of the clinical info sometimes) and
it is selecting it properly.  Will look.

Here is our ovarian instance:
http://sgdi-dev.fhcrc.org/sgdi_ova/

If you go to experiment seleciton and the berchuck2005ESet tab, then click on
the drop-down for 'Debulking'.  There is a blank value (this is likely a bad
input data issue) and if you select it, no samples get selected.  So something
is odd there.