Question about Adjacencies & Inversion

[greg-sharpe]

I have a question about how to specify AdjoiningSequences in an Adjacency in the case of an inversion.

Is it possible to define AdjoiningSequence 2 the same way in each Adjacency, so that AdjoiningSequence 2 could be a single SequenceLocation object referenced by both Adjacencies?

The problem I'm seeing is that the definition of SequenceLocation requires End greater than or equal to Start, meaning that AdjoiningSequence 2 needs to have {Start: C, End: D}. Then it would be impossible to look at either Adjacency and determine that it's describing an inversion, as opposed to describing an adjacency that's present in the reference sequence

(Edited per Larry's correction that End >= Start)

[larrybabb]

In your example above I would refer to two examples that we have images for in the repo currently Reverse complement adjacency example and SV Haplotype example.

in the reverse complement adjacency it illustrates that fact that you only provide the start or the end when specifying either of the SequenceLocations in the adjoinedSequences (aka breakpoint). When you provide a start you are saying that that side of the sequence decreases as it moves away from the adjacency. And likewise when you provide an end you are saying that that side of the adjoined sequence increases as it moves away from the adjacency.

In the Reverse complement adjacency example the 2nd sequenceLocation specifies the end which readily identifies that the second sequence is reverse complemented.

You could apply this to the SV Haplotype example by simply changing the start and end attributes from the 2nd part of the first adjacency with the 1st part of the second adjacency.

@ahwagner please verify my response...

[larrybabb]

I don't think we have "yet" changed our requirement that End >= Start when fully specifying a SequenceLocation. If this is true (and I'll wait on @ahwagner 's validation). That means any inversion would require the use of adjacencies to be defined.

[ahwagner]

In the current model, the Adjacency only describes the sequence junction, nothing beyond. So the use of A and F are not allowed as described in the above model; instead you would have 4 Adjacency objects, ordered in a containing structure (the current proposal is that structure is a revised Haplotype, but there is an open discussion about that at #461). The information for the four adjacency objects, per the current schema and examples would look like:

Adjacency 1: [{start: A}] (if terminal)
Adjacency 2: [{end: B}, {end: D}]
Adjacency 3: [{start: C}, {start: E}]
Adjacency 4: [{end: F}] (if terminal)

Hope this helps.

[greg-sharpe]

Thanks for the clarification!

[Mrinal-Thomas-Epic]

@ahwagner @larrybabb Is the constraint (that only one of start or stop can be specified on a SequenceLocation within an Adjacency specified explicitly anywhere in the schema)?

[ahwagner]

@Mrinal-Thomas-Epic no, there is no constraint specified at the schema level; just the bare-bones work at the moment. I can't get to creating this constraint in the near term, but I think it would be useful to have it. If you agree, please make a separate issue requesting this feature.

[Mrinal-Thomas-Epic]

I agree it would be useful to have. I've created #474. Thanks!