[BDG-FORMATS-29] Re-organize the Feature schema

[tdanford]

This is attempting to re-organize the Feature schema, along the
discussions that we (Timothy, Uri, Matt, Frank) have had in email and on
the phone. The main requirements are:

• less file-format dependence in the field choice ('qValue'-like fields
  could be relegated to the 'attributes' field)
• fewer fields to improve the memory footprint

[tdanford]

This PR reflects a fix for Issue #29

[laserson]

Last time, you mentioned the possibility of multiple accessions in a single database. Does this reflect your resolution of this potential issue?

[tdanford]

Yup! This is meant to be an ID->DB map, but I haven't done any downstream testing yet.

[laserson]

Is it ever possible that an object can have one accession in multiple databases? I don't know if we're talking about a 7-sigma issue here, but is it worth going back to your original proposal of having an array of Dbxref objects?

[tdanford]

Well, and ID->DB (as opposed to DB->ID) mapping would handle "multiple acc's in one DB," right? I'm agnostic either way.

[laserson]

No, I mean, what if the same accession is used in multiple databases? Does
this ever happen?
On Sep 15, 2014 12:39 PM, "Timothy Danford" notifications@github.com
wrote:

In src/main/resources/avro/bdg.avdl:

• union { null, double } signalValue = null;
• union { null, double } pValue = null;
• union { null, double } qValue = null;
• union { null, long } peak = null;

• /**
• The value associated with this feature (if double)
• */
• union { null, double } value = null;

• /**
• Cross-references into other databases.

• Key is database name and value is the accession.
• */
• map dbxrefs = null;

Well, and ID->DB (as opposed to DB->ID) mapping would handle "multiple
acc's in one DB," right? I'm agnostic either way.

—
Reply to this email directly or view it on GitHub
https://github.com/bigdatagenomics/bdg-formats/pull/30/files#r17563834.

[tdanford]

Oh. Um. Yeah, maybe. "7-sigma," like you said, but we should probably plan for it.

[laserson]

Should we go back to your initial suggestion then? Of an array of Dbxref objects?

[laserson]

+1 overall

[fnothaft]

@tdanford would you like to get this in before I cut a new bdg-formats release?

[tdanford]

I don't think it's necessary, no.

[fnothaft]

OK, thanks. I'll cut the release now.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/bdg-formats-prb/44/

[tdanford]

Should this be rebased down before merging?

ANSWER: YES.

Let me do that, real quick.

[fnothaft]

Yes, please squash.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/bdg-formats-prb/45/Test PASSed.

[fnothaft]

Merged! Thanks @tdanford!