VG RDF uses local non standard URIs for its own concepts

[JervenBolleman]

We should mint URIs for the concept of a node in the graph, path, step in path, and before.
i.e. a mini ontology for the following concepts.

Something like this

The following rdfs classes are present in the data (can be left implicit via an RDFS construct as they are not that interesting or needed for likely queries).

A :Node in a Variation Graph is a resource that represents a stretch of continuous DNA. They have an identity in the graph, as well as an associated IUPAC representation of the DNA molecule.

A :Path in the Variation Graph links nodes in the graph into an order (via steps) that represents a linear DNA sequence as found in a single person/assembly.

A :Step belongs to a single :Path and is ordered by the :step predicates value.

The RDF presentation depends on the following predicates

:before is used to link :Nodes together in the directed graph flowing in the general 5' to 3' direction of the consensus assembly. rdfs domain and range are :Node

:step has as domain the :Step type and as a range xsd:positiveInteger for each :Path a step has an unique value.

:node links a :Step in a :Path to the specific :Node. has as domain :Step and range :Node

[ekg]

I think this is a good start. We will also need to represent entities and
paths on the reverse strand. This is necessary to support concise
representations of inversions, translocations, and assembly graphs. So
instead of linking nodes our edges link :Sides, which correspond to a node
end, and each step in a path can be on the forward or reverse complement
(:Traversal maybe). Does this make sense?
On Oct 1, 2015 12:16 PM, "JervenBolleman" notifications@github.com wrote:

We should mint URIs for the concept of a node in the graph, path, step in
path, and before.
i.e. a mini ontology for the following concepts.

Something like this

The following rdfs classes are present in the data (can be left implicit
via an RDFS construct as they are not that interesting or needed for likely
queries).

A :Node in a Variation Graph is a resource that represents a stretch of
continuous DNA. They have an identity in the graph, as well as an
associated IUPAC representation of the DNA molecule.

A :Path in the Variation Graph links nodes in the graph into an order
(via steps) that represents a linear DNA sequence as found in a single
person/assembly.

A :Step belongs to a single :Path and is ordered by the :step predicates
value.

The RDF presentation depends on the following predicates

:before is used to link :Nodes together in the directed graph flowing in
the general 5' to 3' direction of the consensus assembly. rdfs domain and
range are :Node

:step has as domain the :Step type and as a range xsd:positiveInteger for
each :Path a step has an unique value.

:node links a :Step in a :Path to the specific :Node. has as domain :Step
and range :Node

—
Reply to this email directly or view it on GitHub
#109.

[JervenBolleman]

@ekg Something like this

So instead of having a :before predicate we should maybe have

linksOn35to53 i.e. what was :before

linksOn35to53ofComplementOf and use this predicate to say that the end instead of

not attached here
|
v
actgaga
tctcagt
      /\
       | but here

We can then also split the :step predicate into two subproperties, :stepOnForward and :stepOnReverse.

This makes the query example a bit more complicated.

PREFIX :<http://base/> PREFIX rdf:<ttp://www.w3.org/1999/02/22-rdf-syntax-ns#> 
SELECT ?path (group_concat(?sequence; separator='') as ?pathSeq)
WHERE
{?step :path ?path; 
              :node ?node ;
              {
                   :stepOnForward ?order.
                   ?node rdf:value ?sequence
               }  UNION {
                   :stepOnReverse ?order.
                   ?node rdf:value ?toComplement
                   BIND(dna:reverseComplementOf(?toComplement) AS ?sequence)
               }
}
GROUP BY ?path 
ORDER BY ?order"

The dna:reverseComplementOf would be a custom function, or in the very rare case that a SPARQL endpoint does support custom functions a bunch of nested REPLACE operators.

[adamnovak]

I'm going to say that #217 fixed this. This can be re-opened if we want to further improve our URLs.