Extending the Specification

18 views
Skip to first unread message

Michael Milton

unread,
Aug 5, 2024, 3:06:41 AM8/5/24
to ISAforum
I'm interested in adding some domain-specific metadata to the metadata model. For example, I might want to use MINSEQE where the study (investigation?) is a sequencing experiment. Is there a sanctioned way to "extend" the ISA-JSON spec, and if so, how?

On a related note, are you interested in defining official extensions such as this, so that other standards can integrate into ISA?

Michael Milton

unread,
Aug 29, 2024, 8:31:29 PM8/29/24
to ISAforum

Adding Philippe's reply:

***

Hi Michael,

Thanks for reaching out and for the interest.

The topic of refining / extending the ISA specifications is definitely of interest.

So far, the ISA framework relied on ISA configurations to accommodate annotation requirements:

Please see the following repo for examples of how various checklists have been represented using ISA entities

https://github.com/ISA-tools/isa-configurations/tree/development

There are known limitations to the approach:

1. combinatorial effects

2. maintainance and versioning + keeping in synch with originator

3. alignment into the ISA model from external models/representations (SRA, PRIDE, GEO, etc..)

4. making ISA-API object constructors aware of ISA configurations.

As for MINSEQE, the default ISA configuration aligns to SRA 1.5 model and has annotation profiles for RNA-Seq, ChIP-Seq, CNV-Seq, DNAmethylation-SEQ. Updates are needed to take into account more recent evolution of sequencing technology (ATAQ-Seq, single cell sequencing )

We are currently working with several teams, in Germany, UK and the EMBL-EBI to build a brokering service to deposit multimodality studies to EMBL repositories (as part of ELIXIR Biohackathon, taking place in Barcelona in November 2024).

It would be great to hear the needs and have background information about the projects you are supported. 

As I said, I am very interesting in engineering suggestions and needs from the community.

Happy to get on a call to discuss this further.

Cheers

Philippe

Michael Milton

unread,
Aug 29, 2024, 8:31:58 PM8/29/24
to ISAforum
And my own reply:

***

Hi Philippe,

Thanks for the reply, sorry for my own late reply.

Knowing about configurations is helpful. I actually think a repository of community extensions to the spec is a good, flexible design. Are there some docs somewhere that explain how they integrate into the spec? Also, I note that all the config in that repo relates to the ISA-TAB format, but I'm more interested in ISA-JSON. Does it have a similar extension mechanism?

If only one configuration can be used at a time, then I agree that this is a limitation in the spec, and relaxing this would allow you to separately version multiple extensions, which sounds ideal.

Cheers,
Michael

Philippe Rocca-Serra

unread,
Sep 4, 2024, 12:46:16 AM9/4/24
to Michael Milton, ISAforum
Hi Michael,
 
The ISA xml configurations were meant to be used with ISAcreator tool, a Java standalone app.
It was defined with specific xsd.
But this is not so well documented.

Thanks for the background information about your project. It seems you may have missed sections of our documentation about ISA json.
We have a json schema describing ISA
With classes to support the manipulation of objects defined by the model:

We also have several json led context files (OBO foundry based, wiki data based, sio based and we are adding more).
But this still does answer your initial request, which is the ability to extend the core isa schema to  define annotation requirements and make them available to the object manipulation layer.

I’ll following with pointers to 2 projects and we can continue the discussion 

Best wishes

Philippe 

On Tue, 3 Sep 2024 at 07:20, Michael Milton <ttmig...@gmail.com> wrote:
Hi Philippe,

Thanks for these clarifications. The link you provided seems like a relevant use case to me, but I'm struggling to find the corresponding XML configuration files. It seems like the configuration format isn't formally documented anywhere, is that correct?

I'm working on a research data management project where we want to be able to capture file level metadata but also abstract entities like the experiment, the researchers, and the wet lab metadata, and also align with other existing standards where possible. I think it would be great to better support the JSON format using some common standards. For example if ISA-JSON were defined in terms of a JSON schema or JSON LD, that would provide a lot of tooling automatically. From there, it would be ideal if there were a clearly specified way of extending the base ISA specification to add domain specific metadata.

Cheers,
Michael

On Mon, Sep 2, 2024 at 10:07 PM Philippe <procc...@gmail.com> wrote:

Hi Michael,

my turn to apologize 

The ISA configurations can be used to validate either serialization but it is true that validation requires converting ISA-JSON to ISA-TAB.

We have not developed software support to work directly on json object, ie. making ISA objects aware of annotation requirements defined in a configuration.

A clarification regarding "one configuration used at a time". An ISA configuration can actually define many different assay types. So it means that 'several assay specific annotation requirements' may be checked at once.

However, the restriction happens for describing Source and Sample annotation (s_Study file). So If you work specifically on rodents or flies, you may want to have a specific set of annotations to qualify both Source and Samples.

Documentation:

I'll give you that this bit is not necessarily fully documented and is something that requires attention.

This link https://cmso.science/MIACME/v1.0/ may provide additional information.

Finally, I have a request for you:

Could you give us some background about the projects you are involved in and the needs you have to support as well as the challenges you currently face? How did you come across ISA and what are the things you'd like seen prioritized?

Many thanks in advance.

Philippe


On 19/08/2024 08:48, Michael Milton wrote:
Hi Philippe,

Thanks for the reply, sorry for my own late reply.

Knowing about configurations is helpful. I actually think a repository of community extensions to the spec is a good, flexible design. Are there some docs somewhere that explain how they integrate into the spec? Also, I note that all the config in that repo relates to the ISA-TAB format, but I'm more interested in ISA-JSON. Does it have a similar extension mechanism?

If only one configuration can be used at a time, then I agree that this is a limitation in the spec, and relaxing this would allow you to separately version multiple extensions, which sounds ideal.

Cheers,
Michael

On Mon, Aug 5, 2024 at 6:03 PM Philippe <procc...@gmail.com> wrote:

Hi Michael,

Thanks for reaching out and for the interest.

The topic of refining / extending the ISA specifications is definitely of interest.

So far, the ISA framework relied on ISA configurations to accommodate annotation requirements:

Please see the following repo for examples of how various checklists have been represented using ISA entities

https://github.com/ISA-tools/isa-configurations/tree/development

There are known limitations to the approach:

1. combinatorial effects

2. maintainance and versioning + keeping in synch with originator

3. alignment into the ISA model from external models/representations (SRA, PRIDE, GEO, etc..)

4. making ISA-API object constructors aware of ISA configurations.

As for MINSEQE, the default ISA configuration aligns to SRA 1.5 model and has annotation profiles for RNA-Seq, ChIP-Seq, CNV-Seq, DNAmethylation-SEQ. Updates are needed to take into account more recent evolution of sequencing technology (ATAQ-Seq, single cell sequencing )

We are currently working with several teams, in Germany, UK and the EMBL-EBI to build a brokering service to deposit multimodality studies to EMBL repositories (as part of ELIXIR Biohackathon, taking place in Barcelona in November 2024).

It would be great to hear the needs and have background information about the projects you are supported. 

As I said, I am very interesting in engineering suggestions and needs from the community.

Happy to get on a call to discuss this further.

Cheers

Philippe



On 05/08/2024 08:06, Michael Milton wrote:
I'm interested in adding some domain-specific metadata to the metadata model. For example, I might want to use MINSEQE where the study (investigation?) is a sequencing experiment. Is there a sanctioned way to "extend" the ISA-JSON spec, and if so, how?

On a related note, are you interested in defining official extensions such as this, so that other standards can integrate into ISA?
--
--
--
 
You received this message because you are subscribed to the Google
Groups "ISAforum" group.
To post to this group, send email to isaf...@googlegroups.com
To unsubscribe from this group, send email to
isaforum+u...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/isaforum?hl=en-GB
 
Visit the ISA tools website at http://isa-tools.org and http://isacommons.org
---
You received this message because you are subscribed to the Google Groups "ISAforum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isaforum+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/isaforum/1a900a1f-5254-4f37-a25e-a45e87e4b0cdn%40googlegroups.com.
-- 
Please, note that I share my time between 
the University of Oxford and AstraZeneca, 
therefore my responses may be delayed.
-- 

Dr. Philippe Rocca-Serra, PhD

University of Oxford, UK:
Associate Member of Faculty, Dep of Engineering Science;
R&D Group Co-Investigator: https://datareadiness.eng.ox.ac.uk
________________________________________________________________

Our latest papers: 
Machine Actionable Metadata Models and The FAIR Cookbook.
________________________________________________________________

AstraZeneca, Cambridge, UK:
Senior Director of FAIR Data Collaborations;
Data Office R&D, Data Science &AI Unit.

ORCiD: 0000-0001-9853-5668
twitter:@Phil_at_OeRC

--

  







Reply all
Reply to author
Forward
0 new messages