1 files changed, 61 insertions, 0 deletions
diff --git a/topics/genenetwork/add-metadata-to-trait-page.gmi b/topics/genenetwork/add-metadata-to-trait-page.gmi
new file mode 100644
index 0000000..8e2d433
--- /dev/null
+++ b/topics/genenetwork/add-metadata-to-trait-page.gmi
@@ -0,0 +1,61 @@
+# Add Metadata To The Trait Page (RDF)
+
+Fri 30 Sep 2022 11:48:41 EAT
+
+## Introduction
+
+We are migrating the GN2 relational database to a plain text and RDF database.  Matrix-like data (E.g. fetching sample data for a given data) will be stored inside GN.
+
+So far, we are able to convert the sql data to rdf using "dump.scm" defined in:
+
+=> https://github.com/genenetwork/dump-genenetwork-database
+
+## What are we trying to solve?
+
+Data stored in genenetwork resembles a tree. As an example: we have several species; each of these species belong to a group; each group belongs to a "data type"; and each data type belongs to a particular dataset.  The first step: capturing - albeit requiring more refinement - this data in RDF has been achieved using the aformentioned scheme script.
+
+The overall goal is to be ablet to:
+
+* Incrementally replace MySQL queries with RDF.
+
+* Annotating existing data with metadata that does not yet exist in GN2.
+
+## Goals
+
+In the Trait Analysis page, for example:
+
+=> https://genenetwork.org/show_trait?trait_id=1434280_at&dataset=HC_M2_0606_P
+
+and the corresponding GN1 link:
+
+=> http://gn1.genenetwork.org/webqtl/main.py?cmd=show&db=HC_M2_0606_P&probeset=1434280_at
+
+which on further inspection presents metadata on that specific dataset group here:
+
+=> http://gn1.genenetwork.org/webqtl/main.py?FormID=sharinginfo&GN_AccessionId=112
+
+We notice that there's metadata in GN1 - which we have in RDF - that we can add to the GN2 traits page.  As such, this design doc will be limited to using RDF to:
+
+* Append metadata about the tissue
+* Append relevant metadata about the dataset group, in particular: about the data values and it's processing; about the array platform; experiment type; and contributors.
+
+Beyond querying metadata, this design doc also proposes the creation of a monadic rdf-fetch similar to what happens in:
+
+=> https://issues.genenetwork.org/topics/maybe-monad
+
+### Non-goals
+
+* Refactoring base classes/sql to solely use RDF.
+* Using federated queries - they are slow.
+* Writing a script in Guile to fetch and append extra metadata from wikidata and insert them into RDF as extra nodes.  This should be tackled as a separate issue.
+
+## Actual Design
+
+* Rewrite the existing way of fetching RDF using pymonads.
+* React to the change-amplification - should any exist - caused by the above change and add tests where feasible.
+* Create endpoints to add extra annotations for Tissue, Dataset Group, Dataset Values and Processing; array platform; experiment type; and contributors.
+* Add metadata as links, tooltips, or html <summary> tag to the relevant html section(s).
+
+## Resources
+
+=> https://www.linkedin.com/pulse/six-secret-sparql-ninja-tricks-kurt-cagle/