Help:Registry 2.0

Revision as of 15:37, 6 April 2006 by Randy (Talk | contribs)

The Registry web site is now 3.5 years old - it is time for a rewrite !

The current version of the Registry web site provides these functions:

  • Part data
  • User and group accounts
  • Part viewer and editor
  • DNA Repository
  • Other tools (Blast, Subpart search, assembly lines)

Because part data is hidden in the database, other groups cannot build tools that work with these parts. This limits the rate of development and the variety of tools in all these areas. In particular, the user interface to the Registry is set by the Registry. This problem has an obvious solution: Provide part information in an XML format (e.g. SBML).

consists of a mySQL database, server and user interface code written in PERL, and HTML and Javascript that run on the browser. The Registry provides functions of part display, part modification, grouping and display of parts by category such as Terminator or iGEM2006_MIT. The Registry also supports tools such as searching, subpart search, superpart search, assembly lines, BioBrick Blast, DNA Repositories, Wiki-like comments on parts, user accounts and groups.

Because of this design, the information about each part is hidden in the database. This prevents others from making tools that need part data.

Three changes to the Registry that would make adding parts much easier and would make the software of the Registry much simpler. They would remove current functionality, but would let us move the Registry database to NCBI or D-Space. (These changes would be implemented over the summer and we could move the data after the Jamboree) Please start thinking about these issues.


Accession Numbers

We could dramatically simplify Adding Parts to the Registry if we removed the need to select a part name. Instead, we could use Registry-generated accession numbers. We could start with BB_00001 This would eliminate the letters in the part name as well. In the past we have found that the users like to arrange patterns in the part numbers, but I don't know how important that really is now.

No Types or Categories

The second change that would simplify the registry and part management would be the elimination of part types and categories. Instead, we could do a free-text search and build a table of the resulting parts. Of course, the part designer would be able to add key words inside comments or something and we would seed them with things like the group name and current category. We would probably lose the parameter and table formatting we have now (for example RBS strength or terminator efficiency). Instead, we would have a list of parts. All the tables would be the same, with rows like: 

    BB_47867    Two promoter screening plasmid v1.1	Parameter1:      10-100 copies

Obviously, we would loose the part icons since the types are gone, but perhaps we could let the designer specify the icon for each part - even user icons like: Ralphicon.gif

We would also lose software-enforced consistency in parameter names.

No User Groups and Unchanging Parts

The next change that would simplify adding parts would be the elimination of user groups. Instead, we could give out "Temporary Accession Numbers" freely BB_T841878417937, for example. Anyone on earth could have one. However, these would not show up in any table or any tool in the registry unless specified by number. Each one would be given a password that would be required to make changes in the part. We might (or might not) delete these parts after a month or so.

A temporary accession number could be replaced with a permanent accession number after formal application and receipt of the actual DNA for the part and approval. At that point, the part would not be editable by anyone but the registry staff, or perhaps not by anyone. Each such part would have an associated Wiki Page.

Obviously, there would be no way for a lab or iGEM team to see all the parts they are working on until they are formally accepted.

Summary

Summary, these changes would move the contents of the Registry more towards a static resource and away from a set of tools. In my discussions with many people, they think of the Registry in just that way. It is just like Genbank, isn't it? Once we have completed these changes, we can move the Registry information into some RDF/WML format (text not mySQL) and put it on D-Space or NCBI. Then, everyone can make tools as they wish and the current Registry activity can become more of a Google-like search engine.

Randy