>>Is it possible to pass back custom fields (per spot) from a
>>transformational plugin (with names that BASE doesn't recognize) or can
>>that only be done with gene/reporter lists (the "value" field)? Also,
> Yes - you can have a transformation send associated data back to base (e.g.
> for a lowess normalization you could have the correction values for each
> feature sent back). It's just a question of including it in the out file
> (design the plug-in accordingly). These values are then available for
> filtering and will also appear in tables and can be exported along with data
> using the export function (or sent to new down stream plug-ins).
So as long as I give the column some name that hasn't been used yet,
BASE will remember the new column name and include it in the future on,
for example, a new plug-in setup page (for some downstream plug-in that
uses that custom field)? Also, what determines whether new fields fall
under the "most", "much", "interesting", or "important" categories (how
do we specify)?
>>how do you use that "value" field in any useful way in the BASE
>>interface (filters, etc). I can't seem to filter based on that number,
>>only on membership/non-membership in the gene list.
> The value field in the reporter lists can be used to sort genes (e.g. in
> experiment explorer under reporter search). If you select to filter on a
> reporter list here, the value field will appear as a column and this can be
> sorted on).
I'm realizing most of the custom data I'd want to add are actually
per-reporter, not per-spot, so I can't really do the "make up a new
column via some transformation plugin" thing because you can't make
transformation plug-ins when you export data that's merged on the reporters.
I'd probably have to send it back as values in a gene list, and somehow
do something useful with the values. I'm figuring out how to sort
things on the values... but that's about it. It isn't possible to do
any filtering on the values?
The kinds of data I want to send back are plug-in computed numbers like
"mean fold change" or "quantile fold change" or "t-test p value", etc.
I'd preferably like to compute these values on all reporters and use
BASE to do the filtering (as opposed to using the plug-in to filter).
Alternatively I could make a transformation plug-in that returns the
same number for every spot belonging to the same reporter. A bit
wasteful and it confuses the notion of how BASE separates spot-specific
"columns" and reporter-specific "fields".