Describe how to query without reference bases

[mcupak]

The spec requires specifying referenceBases and alternateBases. Some beacons don't store reference bases. Is this a case that should be supported by the spec? If so, should we consider making the parameter optional?

[antbro]

+1

Miro Cupak wrote:

The spec requires specifying |referenceBases| and |alternateBases|.
Some beacons don't store reference bases. Is this a case that should
be supported by the spec? If so, should we consider making the
parameter optional?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#59, or mute the thread
https://github.com/notifications/unsubscribe-auth/AI_EVFL0pGTJnWKnOuuYwRiAkesspWkRks5qZ6KpgaJpZM4JWe8v.

[jrambla]

+1

[mitsuhashi]

+1

Hi Miro,
Thanks for sparing your time at GA4GH 4th plenary meeting.

Nobutaka

[mbaudis]

+1

[sdelatorrep]

+1 to include this in v0.4

[mcupak]

This proposal was discussed during the workshop yesterday and rejected. Reference bases remain a mandatory parameter as it is easy to fetch the reference for beacons that don't store it in the database directly.

[mbaudis]

But not for structural, anything with a considerable size. So no reference bases if one of the "end" parameters is specified. Michael.

…

On 24 May 2017, at 17:48, Miro Cupak ***@***.***> wrote: This proposal was discussed during the workshop yesterday and rejected. Reference bases remain a mandatory parameter as it is easy to fetch the reference for beacons that don't store it in the database directly. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#59 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AApM1rFmOXSwd9xKgCSLalcshaTYC7dsks5r9FFFgaJpZM4JWe8v>.

[antbro]

And what about a 3 base deletion? Where do we draw the line? Could be easier just to make it always optional? Tony Michael Baudis wrote:

…

But not for structural, anything with a considerable size. So no reference bases if one of the "end" parameters is specified. Michael. > On 24 May 2017, at 17:48, Miro Cupak ***@***.***> wrote: > > This proposal was discussed during the workshop yesterday and rejected. Reference bases remain a mandatory parameter as it is easy to fetch the reference for beacons that don't store it in the database directly. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub <#59 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AApM1rFmOXSwd9xKgCSLalcshaTYC7dsks5r9FFFgaJpZM4JWe8v>. > — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#59 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AI_EVJoq9peCC3zzfayo9yEjswudA0xCks5r9FHdgaJpZM4JWe8v>.

[mbaudis]

@antbro I think the "if one of the end" positions makes this explicit. I'm anyway for optional, but this seems a good solution - if you specify an end, you fulfil a specific requirement.

[antbro]

Hi Michael We have to two things to consider... 1) the 'Reference Allele Sequence' and 2) the fuzziness of the region we're searching for it in Whether or not we have to specify the ref allele sequence 'Reference Allele Sequence' is challenged by the fact that it will be easy for SNPs, harder for indels, and impossible for SVs. So my question was how we know when we have to include a 'Reference Allele Sequence' I do not understand why this is influenced by (or guided by) whether or not we are searching within a precise start-end region or a fuzzy region T Michael Baudis wrote:

…

@antbro <https://github.com/antbro> I think the "if one of the end" positions makes this explicit. I'm anyway for optional, but this seems a good solution - if you specify an end, you fulfil a specific requirement. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#59 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AI_EVKXoIC882mdS2K1AKRO1cg-00qGNks5r9GLHgaJpZM4JWe8v>.

[mbaudis]

I do not understand why this is influenced by (or guided by) whether or
not we are searching within a precise start-end region or a fuzzy region

The "if end, then no ref needed..." proposal is independent of fuzziness (aka imprecision ...). It is just "you provide the sequence being altered, OR you give the coordinates".

And if end == start, you solve the reference requirement in a sneaky but explicit way even for single bases.

So, while I understand the specification of reference_bases for single replacements, and understand the current use for small dels (ref >> alt; e.g. ref ATCGTA, alt A == deletion of 5 bases), I'm not particularly fond of it; but there are sensible ways to query around this if implementing the "if not ref than end else fail" paradigm.

(I'm the mediator here, right?)

[mbaudis]

@mcupak Re-opening. There was general consensus before to make reference_bases optional, so this needs explanation.

This proposal was discussed during the workshop yesterday and rejected. Reference bases remain a mandatory parameter as it is easy to fetch the reference for beacons that don't store it in the database directly.

Reference bases have to be optional for imprecise and range queries, and per se only apply to SNV/small indels. So they only are specified when

• an exact position is being queried
• alternate_bases are being provided

In summary this needs a documentation about these conditional interpretations.

[mbaudis]

Closing this, since in documentation now "N" allele documented as permitted value for unknown, which solves this (if interpreted as "any").