Versioning for entries with sequence and annotation

[cybersiddhu]

A versioning model will be applied for majority of the sequence features. The idea will be primarilly borrowed from GenBank. Every feature will have an sequence id(internal and akin to GID or PID in GenBank) and stable identifier(accession no in GenBank). The stable identifier always starts with version 1. Any change in feature sequence will create a new feature entry with a new sequence id whereas the stable identifier remain intact and increment its version number(1 becomes 2). In other words, all features with identical stable identifiers will differ in their version and the one with higher version would be the canonical one. The feature history of sequence changes will also be preserved.

Tied to https://github.com/dictyBase/Migration/issues/5

[cybersiddhu]

Chado database implementation

Data model

• The data model only applies to row/entry in chado feature table.
• They entry should be versioning enabled, because not all entries need versioning. The feature type could be used to distinguish.
• The value of uniquename column will have the unstable identifier equivalent to gid or pid of genbank that will change with change in sequence.
• The accession column of `dbxreftable will have the stable identifier(accession no) and theversion`` column will have the version number.
• Link the feature with stable identier by using dbxref_id column in feature table.

Managing versions

• After a change in sequence, save the new entry from the clone of previous along with the new sequence and a new value for uniquename column. That uniquename value will be the new unstable identifier for the new sequence entry.
• Create a `dbxrefentry by copying theacessionvalue from the previous but increasing the value ofversion`` column(+1).
• Link the new dbxref entry with the new feature entry. This will be the new entry with an increased version no.

[cybersiddhu]

Software implementation

Bootstrap the data model

• Write a post gff3 plugin or a seprate script that will bootstrap versioning for the loaded features.
• The code need to determine the list of features that needed to be versioned and will on act on changing their data model.
• For each entry, create a dbxref entry by copying the uniquename value and adding 1 to the version column.
• Link the features to their corresponding dbxref entries.
• For each entry overwrite their uniquename column with the value of primary key. A uuid value would be another alternative.

Managing versions

Its beyond the scope of this issue, should be handled by the middleware.

Note: This implementation does not bring the versioning schema for the existing sequencing center and curated gene models.