Deseq2 wrapper question
Hi All, we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file Type CTC CTC LM LM PT PT and we want the select list to contain CTC LM PT In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate. Thanks, Rui
Hello Rui, We have DESeq2 wrapper in our oqtans package and you will find this at https://github.com/ratschlab/oqtans_tools/tree/master/DESeq2 I am currently creating the associated tool_dependency.xml file for dumping the package in galaxy toolshed. otherwise it requires some manual effort to install via toolshed. May be it will help you, Vipin Hi All,
we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file
Type CTC CTC LM LM PT PT
and we want the select list to contain
CTC LM PT
In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate.
Thanks, Rui
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
Hi Vipin and Rui, please have at the toolshed and test toolshed before developing wrappers. For deseq2 there are already wrappers that are trying to solve your issues and it would be nice if you could join that effort. Help testing and contributing to it. Ross has a cool wrapper of edgeR, DESeq2 and voom in one tool! http://testtoolshed.g2.bx.psu.edu/view/fubar/differential_count_models There is also that a DESeq2 single wrapper, more simplistic but functional and with full installation of DESeq2, R3 in a reproducible way. http://testtoolshed.g2.bx.psu.edu/view/bgruening/deseq2 Please help Ross and me to make these wrapper rock solid! Bjoern
Hi All,
we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file
Type CTC CTC LM LM PT PT
and we want the select list to contain
CTC LM PT
In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate.
Thanks, Rui
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
Hi Rui, I should provide a 'sample metadata table', a tabular file with one column the names of your samples, the other column the associated metadata (e.g. type). This is how I've done it in my DESeq2 wrapper. :-) Feel free to check the interface at http://toolshed.bits.vib.be/view/joachim/deseq2 Cheers, Joachim Joachim Jacob Contact details: http://www.bits.vib.be/index.php/about/80-team On 09/11/2013 01:15 AM, ruiwang.sz wrote:
Hi All,
we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file
Type CTC CTC LM LM PT PT
and we want the select list to contain
CTC LM PT
In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate.
Thanks, Rui
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
Wow - that makes at least five Deseq2 wrappers for Galaxy available or in progress :( Bjoern's wrapper on the (test) tool shed, http://testtoolshed.g2.bx.psu.edu/view/bgruening/deseq2 Ross' combined wrapper for edgeR, DESeq2 and voom in one tool: http://testtoolshed.g2.bx.psu.edu/view/fubar/differential_count_models Joachim's wrapper on the Tool Shed http://toolshed.bits.vib.be/view/joachim/deseq2 Vipin's which is soon to be released to the (Test?) Tool Shed, https://github.com/ratschlab/oqtans_tools/tree/master/DESeq2 And Rui's is working on one too. This does seem like duplicated effort & a source of confusion for end users and Galaxy administrators (a problem not unique to deseq2, but affecting many Galaxy wrappers). I appreciate there will be different needs, and one wrapper may not suit all, but I would prefer if the default behaviour for Galaxy tool wrapper authors was to collaborate on one good wrapper rather than writing competing ones. When I started work on a new wrapper I tried to announce this on the mailing list to find out if anyone else was already tackling the same tool - and that seemed to work quite well. Of course, the volume of emails on galaxy-dev has grown quite a lot over the last few years so that may not be as effective, but the archives should be searchable. Perhaps we need to improve communication in some way? One option would be to encourage greater use of the Test Tool Shed for works in progress to give them visibility? You could even have the Tool Shed itself require a search step before creating a new repository to avoid accidental duplication of effort? Or maybe a wiki page of wrappers in progress? (And maybe we should split this into a new thread) Regards, Peter On Wed, Sep 11, 2013 at 8:59 AM, Joachim Jacob | VIB | <joachim.jacob@vib.be> wrote:
Hi Rui,
I should provide a 'sample metadata table', a tabular file with one column the names of your samples, the other column the associated metadata (e.g. type).
This is how I've done it in my DESeq2 wrapper. :-) Feel free to check the interface at http://toolshed.bits.vib.be/view/joachim/deseq2
Cheers, Joachim
Joachim Jacob Contact details: http://www.bits.vib.be/index.php/about/80-team
On 09/11/2013 01:15 AM, ruiwang.sz wrote:
Hi All,
we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file
Type CTC CTC LM LM PT PT
and we want the select list to contain
CTC LM PT
In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate.
Thanks, Rui
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
It is not necessarily bad. If we all test each others tools (dependencies, interface, output,...) we might get to the best of all worlds :-) (I have planned to test Björns wrapper) One of the points of differentiation is also how 'integrated' a tool needs to be: e.g. Ross' tool for diff expression (combining a complete pipelin), and my tool just on DE with DESeq2. I like to keep things granular on the level of the tools. Cheers, Joachim Joachim Jacob. Contact details: http://www.bits.vib.be/index.php/about/80-team On 09/11/2013 10:56 AM, Peter Cock wrote:
Wow - that makes at least five Deseq2 wrappers for Galaxy available or in progress :(
Bjoern's wrapper on the (test) tool shed, http://testtoolshed.g2.bx.psu.edu/view/bgruening/deseq2
Ross' combined wrapper for edgeR, DESeq2 and voom in one tool: http://testtoolshed.g2.bx.psu.edu/view/fubar/differential_count_models
Joachim's wrapper on the Tool Shed http://toolshed.bits.vib.be/view/joachim/deseq2
Vipin's which is soon to be released to the (Test?) Tool Shed, https://github.com/ratschlab/oqtans_tools/tree/master/DESeq2
And Rui's is working on one too.
This does seem like duplicated effort & a source of confusion for end users and Galaxy administrators (a problem not unique to deseq2, but affecting many Galaxy wrappers).
I appreciate there will be different needs, and one wrapper may not suit all, but I would prefer if the default behaviour for Galaxy tool wrapper authors was to collaborate on one good wrapper rather than writing competing ones.
When I started work on a new wrapper I tried to announce this on the mailing list to find out if anyone else was already tackling the same tool - and that seemed to work quite well. Of course, the volume of emails on galaxy-dev has grown quite a lot over the last few years so that may not be as effective, but the archives should be searchable.
Perhaps we need to improve communication in some way? One option would be to encourage greater use of the Test Tool Shed for works in progress to give them visibility? You could even have the Tool Shed itself require a search step before creating a new repository to avoid accidental duplication of effort? Or maybe a wiki page of wrappers in progress?
(And maybe we should split this into a new thread)
Regards,
Peter
On Wed, Sep 11, 2013 at 8:59 AM, Joachim Jacob | VIB | <joachim.jacob@vib.be> wrote:
Hi Rui,
I should provide a 'sample metadata table', a tabular file with one column the names of your samples, the other column the associated metadata (e.g. type).
This is how I've done it in my DESeq2 wrapper. :-) Feel free to check the interface at http://toolshed.bits.vib.be/view/joachim/deseq2
Cheers, Joachim
Joachim Jacob Contact details: http://www.bits.vib.be/index.php/about/80-team
On 09/11/2013 01:15 AM, ruiwang.sz wrote:
Hi All,
we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file
Type CTC CTC LM LM PT PT
and we want the select list to contain
CTC LM PT
In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate.
Thanks, Rui
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
On Wed, Sep 11, 2013 at 10:12 AM, Joachim Jacob | VIB | <joachim.jacob@vib.be> wrote:
It is not necessarily bad. If we all test each others tools (dependencies, interface, output,...) we might get to the best of all worlds :-) (I have planned to test Björns wrapper)
One of the points of differentiation is also how 'integrated' a tool needs to be: e.g. Ross' tool for diff expression (combining a complete pipelin), and my tool just on DE with DESeq2. I like to keep things granular on the level of the tools.
That is a good reason for different wrappers, one for the tool itself, and one for a larger pipeline using the same tool. And both can declare a binary package dependency via the Tool Shed using the binary dependency package Bjoern is working on, which I assume will be released on the Test & Main Tool Shed under the shared IUC account once ready: http://testtoolshed.g2.bx.psu.edu/view/bgruening/package_deseq2_1_0_17 Regards, Peter
Am Mittwoch, den 11.09.2013, 10:21 +0100 schrieb Peter Cock:
On Wed, Sep 11, 2013 at 10:12 AM, Joachim Jacob | VIB | <joachim.jacob@vib.be> wrote:
It is not necessarily bad. If we all test each others tools (dependencies, interface, output,...) we might get to the best of all worlds :-) (I have planned to test Björns wrapper)
One of the points of differentiation is also how 'integrated' a tool needs to be: e.g. Ross' tool for diff expression (combining a complete pipelin), and my tool just on DE with DESeq2. I like to keep things granular on the level of the tools.
That is a good reason for different wrappers, one for the tool itself, and one for a larger pipeline using the same tool. And both can declare a binary package dependency via the Tool Shed using the binary dependency package Bjoern is working on, which I assume will be released on the Test & Main Tool Shed under the shared IUC account once ready:
http://testtoolshed.g2.bx.psu.edu/view/bgruening/package_deseq2_1_0_17
Yes that is right, at the moment I have patches for the toolshed in preparation for a unified and easy installation of R dependencies, once that is done it will be under the IUC account. I don't see the need for 5 different wrappers. Ross approach deserves an own wrapper I think. I thought a lot about if we should use it, since it can compare 3 different approaches for DE, but it comes with some complexity. We spend some time in trying to get the UI correct and investigated some possibilities. Currently we have a few guys testing it and giving comments to our UI idea. If the ownershop is a problem, we probably can move the wrappers to a different account. But please lets merge the code base and start communicating about the UI. Thanks, Bjoern
Regards,
Peter
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
Yes I am agreeing on all statements. Let me finish my package and will add in one of the toolshed by today evening. I also willing to test other wrappers to improve the additional parameter settings. Tool wrappers Work In Progress from my side: DEXSeq edgeR Scripture STAR aligner SAFT multi mapper resolution This will be adding soon to the toolshed. Thanks to all, Vipin On Wed, Sep 11, 2013 at 6:08 AM, Bjoern Gruening <bjoern.gruening@gmail.com>wrote:
On Wed, Sep 11, 2013 at 10:12 AM, Joachim Jacob | VIB | <joachim.jacob@vib.be> wrote:
It is not necessarily bad. If we all test each others tools (dependencies, interface, output,...) we might get to the best of all worlds :-) (I have planned to test Björns wrapper)
One of the points of differentiation is also how 'integrated' a tool needs to be: e.g. Ross' tool for diff expression (combining a complete
and my tool just on DE with DESeq2. I like to keep things granular on
Am Mittwoch, den 11.09.2013, 10:21 +0100 schrieb Peter Cock: pipelin), the
level of the tools.
That is a good reason for different wrappers, one for the tool itself, and one for a larger pipeline using the same tool. And both can declare a binary package dependency via the Tool Shed using the binary dependency package Bjoern is working on, which I assume will be released on the Test & Main Tool Shed under the shared IUC account once ready:
http://testtoolshed.g2.bx.psu.edu/view/bgruening/package_deseq2_1_0_17
Yes that is right, at the moment I have patches for the toolshed in preparation for a unified and easy installation of R dependencies, once that is done it will be under the IUC account.
I don't see the need for 5 different wrappers. Ross approach deserves an own wrapper I think. I thought a lot about if we should use it, since it can compare 3 different approaches for DE, but it comes with some complexity. We spend some time in trying to get the UI correct and investigated some possibilities. Currently we have a few guys testing it and giving comments to our UI idea. If the ownershop is a problem, we probably can move the wrappers to a different account. But please lets merge the code base and start communicating about the UI.
Thanks, Bjoern
Regards,
Peter
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
STAR aligner
I have a partial STAR wrapper that may be useful: https://bitbucket.org/jgoecks/jeremys-code/src/d7096c132657420f08b540c6954b9... Whether you use this code or not, I do have one request: please put the output in BAM rather than SAM format in order to save disk space. Best, J.
Vipin, Search works pretty well on the toolshed - I'd recommend a search for (eg) 'star' before writing yet another rnastar toolshed wrapper and dependency package? There's an installable fubar package_rnastar and an rnastar wrapper based on Jeremy's code mentioned below - but working AFAIK - at least it installs and works fine here and on the GVL. On Thu, Sep 12, 2013 at 1:49 AM, Jeremy Goecks <jeremy.goecks@emory.edu>wrote:
STAR aligner
I have a partial STAR wrapper that may be useful: https://bitbucket.org/jgoecks/jeremys-code/src/d7096c132657420f08b540c6954b9...
Whether you use this code or not, I do have one request: please put the output in BAM rather than SAM format in order to save disk space.
Best, J.
Hey Guys, It's great to receive so many responses! Thank you all for the information... and let me check out each of the masterpiece. :-) Best, Rui On Wed, Sep 11, 2013 at 1:56 AM, Peter Cock <p.j.a.cock@googlemail.com>wrote:
Wow - that makes at least five Deseq2 wrappers for Galaxy available or in progress :(
Bjoern's wrapper on the (test) tool shed, http://testtoolshed.g2.bx.psu.edu/view/bgruening/deseq2
Ross' combined wrapper for edgeR, DESeq2 and voom in one tool: http://testtoolshed.g2.bx.psu.edu/view/fubar/differential_count_models
Joachim's wrapper on the Tool Shed http://toolshed.bits.vib.be/view/joachim/deseq2
Vipin's which is soon to be released to the (Test?) Tool Shed, https://github.com/ratschlab/oqtans_tools/tree/master/DESeq2
And Rui's is working on one too.
This does seem like duplicated effort & a source of confusion for end users and Galaxy administrators (a problem not unique to deseq2, but affecting many Galaxy wrappers).
I appreciate there will be different needs, and one wrapper may not suit all, but I would prefer if the default behaviour for Galaxy tool wrapper authors was to collaborate on one good wrapper rather than writing competing ones.
When I started work on a new wrapper I tried to announce this on the mailing list to find out if anyone else was already tackling the same tool - and that seemed to work quite well. Of course, the volume of emails on galaxy-dev has grown quite a lot over the last few years so that may not be as effective, but the archives should be searchable.
Perhaps we need to improve communication in some way? One option would be to encourage greater use of the Test Tool Shed for works in progress to give them visibility? You could even have the Tool Shed itself require a search step before creating a new repository to avoid accidental duplication of effort? Or maybe a wiki page of wrappers in progress?
(And maybe we should split this into a new thread)
Regards,
Peter
On Wed, Sep 11, 2013 at 8:59 AM, Joachim Jacob | VIB | <joachim.jacob@vib.be> wrote:
Hi Rui,
I should provide a 'sample metadata table', a tabular file with one
column > the names of your samples, the other column the associated metadata (e.g. > type). > > This is how I've done it in my DESeq2 wrapper. :-) Feel free to check the > interface at http://toolshed.bits.vib.be/view/joachim/deseq2 > > > Cheers, > Joachim > > Joachim Jacob > Contact details: http://www.bits.vib.be/index.php/about/80-team > > > > On 09/11/2013 01:15 AM, ruiwang.sz wrote: >> >> Hi All, >> >> we are making a wrapper for deseq2, and there is a step where >> control/experimental conditions >> need to be determined. For example, it would be a column like the >> following in the input file >> >> Type >> CTC >> CTC >> LM >> LM >> PT >> PT >> >> and we want the select list to contain >> >> CTC >> LM >> PT >> >> In other words, we want to filter out the 'Type' and have only distinct >> ones in the select list. We could >> make it in two steps, generate a intermediate file for this but it would >> be nice that we could directly >> retrieve the distinct types to make the list. I took a look at the tool >> config wiki, but didn't see anything >> (maybe because it was a quick scan). Does anyone have any tip on this? >> we'll really appreciate. >> >> Thanks, >> Rui >> >> >> ___________________________________________________________ >> Please keep all replies on the list by using "reply all" >> in your mail client. To manage your subscriptions to this >> and other Galaxy lists, please use the interface at: >> http://lists.bx.psu.edu/ >> >> To search Galaxy mailing lists use the unified search at: >> http://galaxyproject.org/search/mailinglists/ > > > ___________________________________________________________ > Please keep all replies on the list by using "reply all" > in your mail client. To manage your subscriptions to this > and other Galaxy lists, please use the interface at: > http://lists.bx.psu.edu/ > > To search Galaxy mailing lists use the unified search at: > http://galaxyproject.org/search/mailinglists/
Hi Joachim, Thanks for the info! but I could not register on your instance...and thus could not check your wrapper. :-( Please help! Thanks, Rui On Wed, Sep 11, 2013 at 12:59 AM, Joachim Jacob | VIB | < joachim.jacob@vib.be> wrote:
Hi Rui,
I should provide a 'sample metadata table', a tabular file with one column the names of your samples, the other column the associated metadata (e.g. type).
This is how I've done it in my DESeq2 wrapper. :-) Feel free to check the interface at http://toolshed.bits.vib.be/**view/joachim/deseq2<http://toolshed.bits.vib.be/view/joachim/deseq2>
Cheers, Joachim
Joachim Jacob Contact details: http://www.bits.vib.be/index.**php/about/80-team<http://www.bits.vib.be/index.php/about/80-team>
On 09/11/2013 01:15 AM, ruiwang.sz wrote:
Hi All,
we are making a wrapper for deseq2, and there is a step where control/experimental conditions need to be determined. For example, it would be a column like the following in the input file
Type CTC CTC LM LM PT PT
and we want the select list to contain
CTC LM PT
In other words, we want to filter out the 'Type' and have only distinct ones in the select list. We could make it in two steps, generate a intermediate file for this but it would be nice that we could directly retrieve the distinct types to make the list. I took a look at the tool config wiki, but didn't see anything (maybe because it was a quick scan). Does anyone have any tip on this? we'll really appreciate.
Thanks, Rui
______________________________**_____________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/**search/mailinglists/<http://galaxyproject.org/search/mailinglists/>
participants (8)
-
Bjoern Gruening -
Björn Grüning -
Jeremy Goecks -
Joachim Jacob | VIB | -
Peter Cock -
Ross -
ruiwang.sz -
Vipin TS