Discussion on priority-feature-requests

[Gaj Stan (BIGCAT)]

Hello Nathan,

A few more feature requests/bugs while working with GO-Elite 1.19.
Some may have already been addressed:

1) One has to find back the correct directory all the time, when
choosing files within the same analysis (and between analyses)

2) Opening screen warns to 'press quit' upon trying to click 'X', but
there is no quit button

3) Permutation should give progress as well. This is by far longest
step, and you have no clue how many have been done so far (and as such
how long it is going to take)

4) Results are NOT stable (resulting GO IDs), small changes are
present between runs on the same data set (I'm looking further into
this and will make a seperate thread about that).

5) You cannot see which file you are currently opening (input,
denominator or output), it would be great if the open dialog told you
so (to prevent errors)

6) Output: the combined files (combined-results, combined-gene-
associations, combined-gene-ranks) do NOT contain the title of your
input file, as all other result files do

7) Once the GO-Elite GUI is minimized, GO-Elite does nothing (CPU: 0%,
Windows XP SP3). Reopening the GUI from the taskbar doesn't work
either. This can get annoying, especially if more than 10 sets need to
be processed. This is probably related to the Python GUI and it's not
something you can easily address. I've been able to replicate this on
both Windows XP notebooks and desktops.

Best wishes,

-- Stan

[Nathan Salomonis]

Hey Stan,

Thanks for the reporting these issues. Most of these still
do need to be addressed. As always, some are more
difficult than others but will be address through the
feature request page. Also, a new webservice for GO-Elite
is available from http://genmapp.org/go_elite.

> A few more feature requests/bugs while working with
>GO-Elite 1.19.
> Some may have already been addressed:
>
> 1) One has to find back the correct directory all the
>time, when
> choosing files within the same analysis (and between
>analyses)

---I will add this to my to do list

> 2) Opening screen warns to 'press quit' upon trying to
>click 'X', but
> there is no quit button

---I will have the 'X' apply a program exit in this place.

> 3) Permutation should give progress as well. This is by
>far longest
> step, and you have no clue how many have been done so
>far (and as such
> how long it is going to take)

---There's a way to do this, but I haven't figured this
out well yet. I will look into it.

> 4) Results are NOT stable (resulting GO IDs), small
>changes are
> present between runs on the same data set (I'm looking
>further into
> this and will make a seperate thread about that).

---This is due to the fact that permutations are random
and thus you can see variation in permute p values from
run to run. This is the case for other permutation based
programs.

> 5) You cannot see which file you are currently opening
>(input,
> denominator or output), it would be great if the open
>dialog told you
> so (to prevent errors)

---This should be doable.

> 6) Output: the combined files (combined-results,
>combined-gene-
> associations, combined-gene-ranks) do NOT contain the
>title of your
> input file, as all other result files do

---Correct, since this can contain many input files with
different names. Thus the input name is stored in the file
in the column of the left. This has been debated a bit.

> 7) Once the GO-Elite GUI is minimized, GO-Elite does
>nothing (CPU: 0%,
> Windows XP SP3). Reopening the GUI from the taskbar
>doesn't work
> either. This can get annoying, especially if more than
>10 sets need to
> be processed. This is probably related to the Python GUI
>and it's not
> something you can easily address. I've been able to
>replicate this on
> both Windows XP notebooks and desktops.

---I didn't know about this one. I will try to test this
myself and talk to a Python guru about it. These should
all hopefully be addressed before the version 1.2 release.
This should also be the official non-beta release version
once the paper is accepted and this version is thoroughly
tested.


> Best wishes,
>
> -- Stan
> >