[Galaxy-user] ideas from using Galaxy2

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

[Galaxy-user] ideas from using Galaxy2

Belinda M. Giardine
While I was trying to do an example showing Galaxy2, I found myself
looking for tools that weren't available.  When working with the mutations
it would be nice to have some function for getting exact matches,
possibly even using the name field.  When the data is mixed with very
large and very small regions just overlap isn't always to most
informative.  Mutations also give you up to 3 names for a single nt
region.  It would take incorporating the name field to distinguish these.

Also when it comes to looking at the queries there is nothing for
displaying more than 1 at a time.  It would be nice to be able to view
them as multiple user tracks at UCSC.  Also graphs that show more than 1
query?  I would have liked somehow either graph or at least a number in a
column to be able to get a count of the number of regions that overlapped
each region in a second query.  Similar to coverage but with counts.

Belinda


Reply | Threaded
Open this post in threaded view
|

Re: ideas from using Galaxy2

James Taylor
> While I was trying to do an example showing Galaxy2, I found myself
> looking for tools that weren't available.  When working with the
> mutations
> it would be nice to have some function for getting exact matches,
> possibly even using the name field.  When the data is mixed with very
> large and very small regions just overlap isn't always to most
> informative.  Mutations also give you up to 3 names for a single nt
> region.  It would take incorporating the name field to distinguish
> these.

I'm not sure exactly what you mean. Join with the option of matching
additional fields? Some kind of filter / match tool that would allow a
variable number of input queries could be quite useful. Perhaps the
easiest thing to do would be to make a script that does what you want,
then integrate it into Galaxy.

> Also when it comes to looking at the queries there is nothing for
> displaying more than 1 at a time.  It would be nice to be able to view
> them as multiple user tracks at UCSC.

Yes, a "build UCSC custom track" tool would be awesome. All of these
tools could be built using the multi-page "wizard" tool functionality
that is in the framework but not used by many tools. Some modifications
would be needed, but it should be straightforward. The idea would be
that a tool could provide a "wizard controller" class that would
evaluate the current tool state and return the appropriate
configuration for the next page or a signal to dispatch to the tool
runner. Most of the pieces are already in place.

> Also graphs that show more than 1
> query?  I would have liked somehow either graph or at least a number
> in a
> column to be able to get a count of the number of regions that
> overlapped
> each region in a second query.  Similar to coverage but with counts.

Yes, we've been talking about a tool for progressively building
complicated plots for a while. Bob recently expressed some interest in
working on this. This could also be a wizard, the fist page would have
general plot configuration and from there you would have the option of
wither "add another line [pointset, whatever]" or "execute/finish".


Reply | Threaded
Open this post in threaded view
|

Re: ideas from using Galaxy2

Belinda M. Giardine


On Mon, 23 Jan 2006, James Taylor wrote:

> > While I was trying to do an example showing Galaxy2, I found myself
> > looking for tools that weren't available.  When working with the
> > mutations
> > it would be nice to have some function for getting exact matches,
> > possibly even using the name field.  When the data is mixed with very
> > large and very small regions just overlap isn't always to most
> > informative.  Mutations also give you up to 3 names for a single nt
> > region.  It would take incorporating the name field to distinguish
> > these.
>
> I'm not sure exactly what you mean. Join with the option of matching
> additional fields?

Pretty much yes.  The scenerio was that I had done overlap with mutations
and some other query getting back the mutations.  I had also done the same
thing with a different second query.  Then what I wondered was how many
mutations were in both results.  I couldn't use overlap since large
deletions overlap many other mutations.  Even if there was an overlap with
exact match option that wouldn't give me the answer I was looking for if
any of the mutations had the same coordinates (ie A>G, A>C, A>T each
listed as a different mutation).

Belinda


 > Some kind of filter / match tool that would allow a

> variable number of input queries could be quite useful. Perhaps the
> easiest thing to do would be to make a script that does what you want,
> then integrate it into Galaxy.
>
> > Also when it comes to looking at the queries there is nothing for
> > displaying more than 1 at a time.  It would be nice to be able to view
> > them as multiple user tracks at UCSC.
>
> Yes, a "build UCSC custom track" tool would be awesome. All of these
> tools could be built using the multi-page "wizard" tool functionality
> that is in the framework but not used by many tools. Some modifications
> would be needed, but it should be straightforward. The idea would be
> that a tool could provide a "wizard controller" class that would
> evaluate the current tool state and return the appropriate
> configuration for the next page or a signal to dispatch to the tool
> runner. Most of the pieces are already in place.
>
> > Also graphs that show more than 1
> > query?  I would have liked somehow either graph or at least a number
> > in a
> > column to be able to get a count of the number of regions that
> > overlapped
> > each region in a second query.  Similar to coverage but with counts.
>
> Yes, we've been talking about a tool for progressively building
> complicated plots for a while. Bob recently expressed some interest in
> working on this. This could also be a wizard, the fist page would have
> general plot configuration and from there you would have the option of
> wither "add another line [pointset, whatever]" or "execute/finish".
>