Add Biomedical Metadata block

[eaquigley]

---

Author Name: Steve Kraffmiller (@sekmiller)
Original Redmine Issue: 3479, https://redmine.hmdc.harvard.edu/issues/3479
Original Date: 2014-02-03
Original Assignee: Kevin Condon

---

Based on general fields in ISA-Tab and NPG's Scientific Data. Working with Stem Cell Commons and other biomedical researchers to develop a list of controlled vocabulary values based on biomedical ontologies (e.g., OBI, etc).

Need to consider support:

1. "type ahead" instead of dropdown if there is a long controlled vocabulary list OR
2. hierarchical terms list rather than flat list
3. List of commonly used organisms according to NCBI in molecular research http://www.ncbi.nlm.nih.gov/Taxonomy/taxonomyhome.html/
4. Long-term:Export/Import w/ ISA-Tab metadata from Stem Cell Commons
5. Long-term: SKOS Concept Scheme: {Tim Clark: for the domain of interest (let's say Biology, or even a subdomain of biology); select the ontology terms from each ontology for each category you have, like "Study Design", and create corresponding SKOS terms under the category, with the relation "isa" from the SKOS term to the corresponding OWL class. THe names of the SKOS terms should be identical to the OWL classes. }

---

Related issue(s): #485
Redmine related issue(s): 3900

---

[eaquigley]

---

Original Redmine Comment
Author Name: Eleni Castro (@posixeleni)
Original Date: 2014-03-25T19:45:51Z

---

From: Eleni Castro ecastro@fas.harvard.edu
Date: Tue, 25 Mar 2014 13:56:37 -0400
To: Tim Clark at HARVARD tim_clark@harvard.edu
Cc: Sudeshna Das sudeshna.das2@gmail.com, "Quigley, Elizabeth" equigley@iq.harvard.edu, "Durand, Gustavo" gdurand@iq.harvard.edu
Subject: Re: Help Reviewing Sample Biomedical Metadata Dataset in Dataverse 4.0 UI

Thanks, Tim. I was actually worried about exactly that. We can support a simple flat dropdown list at the moment but with Gustavo's help need to look into supporting "type ahead" and hierarchical terms, especially if we have a lot of terms and support for multiple ontologies.

For the ID and Accession #: Working on SKOS mapping is something I've never done before but am willing to give it a shot. Thanks for offering to help too (I will probably take you up on that!).

Additionally, when's Sudeshna coming back?

Cheers,

Eleni

Eleni Castro
Research Coordinator, Data Acquisition and Archiving, Data Science
IQSS, Harvard University
http://www.iq.harvard.edu/people/eleni-castro

From: Tim Clark at HARVARD tim_clark@harvard.edu
Date: Tue, 25 Mar 2014 12:50:54 -0400
To: Eleni Castro ecastro@fas.harvard.edu
Cc: Sudeshna Das sudeshna.das2@gmail.com, "Quigley, Elizabeth" equigley@iq.harvard.edu
Subject: Re: Help Reviewing Sample Biomedical Metadata Dataset in Dataverse 4.0 UI

Very sorry to say - this is not usable in any way. No researcher will ever give a term accession number or ontology ID. Never happen. They won't know what you are talking about, or have a clue how to discover the info.

They need a drop-down or type-ahead list. Ask Sudeshna to show you what they implemented for Stem Cell Common, when she returns from India. Do it like that.

Now, you say, but how do I get the ontology ID and accession number? Well, that is why I pointed out the SKOS mapping approach. That is 100% the way to go.

You want to build a SKOS Concept Scheme for the domain of interest (let's say Biology, or even a subdomain of biology); select the ontology terms from each ontology for each category you have, like "Study Design", and create corresponding SKOS terms under the category, with the relation "isa" from the SKOS term to the corresponding OWL class. THe names of the SKOS terms should be identical to the OWL classes.

Then whenever you need the pulldown you have a nice standardized file you can read and parse to get the terms for each category, the terms map to the ontologies behind the scenes, and you can (should) make the SKOS file available read-only to the world via a purl.

There is some detail in the attached PDF. Protege ontology editor should support doing what you need to do.

I can help you do this when I return. If you need help more quickly you could ask Sudeshna when she returns from India. Katie Wolstendroft also would know how to do this. Implementing the SKOS way has many benefits...

Best

Tim

On Mar 25, 2014, at 12:24 PM, Castro, Eleni ecastro@fas.harvard.edu wrote:

Hi Tim and Sudeshna,

Based on our biomedical metadata discussions/meetings I have put together a sample Dataset in our test instance of Dataverse 4.0. We have a biomedical metadata block (see screenshot) that I would like for you to review and let me know if this is how it should look in the form or if we need to adjust (note: we're still working to support grandparent-parent-children elements so that's why some of them seem jumbled together). I'm including a screenshot below of what it currently resembles but I desperately need your help to make sure this is usable for your researchers. If you have some time in the next week or so Elizabeth (our Usability Specialist) and I would love to review this with you.

http://dvn-alpha.hmdc.harvard.edu/dataset.xhtml?id=128

<81FF8753-6C3A-4BCC-B120-4848902FF0D5.png>

Thanks eversomuch!

Eleni

Eleni Castro
Research Coordinator, Data Acquisition and Archiving, Data Science
IQSS, Harvard University
http://www.iq.harvard.edu/people/eleni-castro

[eaquigley]

---

Original Redmine Comment
Author Name: Eleni Castro (@posixeleni)
Original Date: 2014-04-29T17:38:40Z

---

We will have a biomedical metadata block for beta but it will be limited to Stem Cell Commons controlled vocabulary at the moment. This will hopefully be helpful to get feedback from others within biomed on how we can expand support for their particular data. All biomedical metadata for beta has been checked in to git and had the schema.xml updated.

[eaquigley]

---

Original Redmine Comment
Author Name: Eleni Castro (@posixeleni)
Original Date: 2014-05-02T20:07:33Z

---

Might need to create a new ticket for new changes to the Biomedical block but this is ready for beta. See Issue #3900

[eaquigley]

---

Original Redmine Comment
Author Name: Kevin Condon (@kcondon)
Original Date: 2014-05-06T23:46:07Z

---

Tested on 5/6

Works, closing ticket.