Recent changes to bugs

E. coli will not build with gmb 2.0 alpha 6

Jeffrey — Fri, 18 May 2007 04:15:40 -0000

I started the E. coli build last night around 5:30 and this morning it still hadn't finished; I'm attaching a screen shot of the console screen for potential error messages you can decode.

Gene Ontology export not working

Jeffrey — Fri, 18 May 2007 04:13:17 -0000

Gene Ontology tables were messed up in gmb 2.0 alpha 6.

The GeneOntology table was empty, the GeneOntologyCount and Tree tables only had three records in them, the top three nodes.

The alpha5 version of GM builder had created full tables, so I don't know what happened here.

Tallies menu confusing

Jeffrey — Tue, 23 Jan 2007 14:55:48 -0000

The menu items are a little confusing--there is no way to tally the GO entries in the postgres database without doing the "full monty" is that right?

Tallies not correct for Obo dbxref

Jeffrey — Tue, 23 Jan 2007 14:53:25 -0000

I tried out the tally engine for UniProt XML on my large file and it gave me the right number of entries for UniProt, but it was off by 20 for obo/term/def/dbxref, see attached screen shot.

Tallies not correct

Jeffrey — Fri, 12 Jan 2007 06:24:22 -0000

Some, but not all, of the database numbers appear to be exactly double the XML numbers; might you have some prior import sitting in the database? If this was not the case, then we'll have to look at the database queries that you used; perhaps there's a join sitting in there that is incorrectly inflating the number of rows.

John David N. Dionisio, PhD
Assistant Professor, Computer Science
Loyola Marymount University

Arabidopsis not exported correctly

Jeffrey — Fri, 12 Jan 2007 06:20:20 -0000

I have attached a word document that contains just one entry from the Arabidopsis XML to illustrate what I am talking about.

--Some earlier version of GenMAPP Builder was able to detect when a gene ID system used by GenMAPP exists in the XML. The current GM builder can do this to some extent because this is how it is detecting the presence of UniGene as a system (UniGene didn't exist for E. coli). The UniGene part of the XML is highlighted in magenta in the word document. In the Systems table of the access gene database, UniGene has the system code of "U". Right now, GMbuilder is building a UniGene table and the appropriate relationship tables. It is adding a date to the date record of the Systems table for UniGene.

--Similarly, GMBuilder should detect the presence of TAIR as a gene ID system already known about by GenMAPP. TAIR is already listed in the Systems table of the access template file with the SystemCode of "A". The place where you should be getting the TAIR IDs from in the XML is highlighted in green. You should make the TAIR table from there, put a date in the date field of the TAIR record in the Systems table, and change the Columns field to "ID|".

--DO NOTHING with OrderedLocusNames. I have highlighted that part of the XML as yellow. I do not want us to take the IDs from there at all. I don't want the TAIR table derived from these data at all. The OrderedLocusNames field in the Systems table should be left alone, no date put in it. Note this is not a matter of only using OrderedLocus or only using ORF from the XML; I am saying don't take data from this part of the XML at all and do nothing with the OrderedLocusNames field in the Systems table.

This is something specifically to do with Arabidopsis. I have examined the IDs coming from the OrderedLocusNames portion of the XML and have found some problems. Since the data are available in another part of the XML record, I would rather take TAIR IDs from there than from the OrderdLocusNames portion of the XML.

Also, in the Relations table, there are System codes being used there that are incorrect. The Systems table is the place to look for the official System codes. I don't know how the incorrect codes cropped in because this wasn't an issue with the E. coli database.

Thanks,
Kam

Dist script needs to be more robust

dondi — Mon, 24 Jul 2006 07:05:21 -0000

The gmbuilder.bat script that is currently included with GenMAPP Builder
assumes that java is on the user's path. Kam and I discovered that the
latest Java installer does not actually do this for you; what can we do to
handle this situation?

If we can't build this into the batch file, I can think of a few alternatives:
- Provide a README in the dist that tells the user how to add Java to their
path.

- Rearrange the dist so that gmbuilder.jar is at the top of the directory,
with supporting libraries in lib/. The user then double-clicks on
gmbuilder.jar to run. gmbuilder.bat can be used for a more advanced
user who would like to tweak the heap memory settings or modify the log
level.

- As a variation on the previous option, allow for some optional
properties file (e.g., gmbuilder.properties) to live in an expected location;
this properties file can then be edited by the user to change the heap
memory settings, log level, or whatever else. This way, starting GenMAPP
Builder remains standardized on the action of double-clicking on
gmbuilder.jar.

Ah, the things we learn when we start downloading and installing our
software on different machines :)

IDs that take the form “ECOK12F050”

Joey J. Barrett — Mon, 26 Jun 2006 23:35:52 -0000

Kam,

Please notify me what you would like done with IDs that
take the form “ECOK12F050”.

Thanks,
Joey

Blattner IDs of the form "bxxxx/bxxxx"

Joey J. Barrett — Sun, 25 Jun 2006 20:47:07 -0000

Some Blattner IDs are of the form "bxxxx/bxxxx". These
must be split into two entries in the table where the
other column associated with the entry are replicated
for each ID.

X-table to Blattner are not being constructed

Joey J. Barrett — Sun, 25 Jun 2006 20:43:33 -0000

The X-table to Blattner tables are not being
constructed. There is a bug in the species profile
relationship table generation and should be taken care
of immediately.

This does not throw an exception but rather it is an
error of validity from what the user chooses to the end
product.