[GUSDEV] SequenceGroup table

[Chris S. <sto...@pc...>]

Hi,
We are loading a new set of OrthoMCL generated ortholog groups into  
an instance of GUS and found that the current  SequenceGroup parent  
table has several specifically-named attributes (i.e., no generic  
attributes) that we can use for our purposes without renaming. The  
attributes we want to capture are quality metrics for an ortholog  
group. For example we'd like to capture the ave e-value rather that  
min and max as currently provided in SequenceGroup and all of its  
views including OrthologGroup.

I'd like to make the following proposal, get feedback, and put in the  
GUS schema bug tracker.
1) Create a OrthoMCL view. This would be specific to our GUS  
instances but could be used as a template for renaming attributes in  
OrthologGroup. On the right I've indicated with <- the parent  
attribute that might be used where the name differs.

Column	nulls?	type
SEQUENCE_GROUP_ID	NO	NUMBER
SUBCLASS_VIEW	NO	STRING(100)
NAME		STRING(500)
DESCRIPTION		STRING(2000)
NUMBER_OF_MEMBERS	NO	NUMBER
NUMBER_OF_TAXA		NUMBER
AVE_PERCENT_IDENTITY		FLOAT <-MIN_MATCH_IDENTITY
AVE_PERCENT_COVERAGE		FLOAT  <- MIN_PERCENT_MATCH
AVE_E_VALUE		FLOAT <- MIN_PVALUE_MANT
AVE_DOMAIN_SHARING		FLOAT <- MAX_SCORE
BLAST_MATCH_RATIO		FLOAT <- MIN_SCORE
SEQ_GROUP_EXPERIMENT_ID		FOREIGN KEY
plus the usual housekeeping columns

In this proposal, the OrthoMCLGroup attributes would have the same  
type but different semantics as implied by the names of attributes in  
the parent class and queries can easily avoid this by use of subclass  
view.

2) Rename the SequenceGroup attributes to be more generic. Suggested  
changes are:
MAX_MATCH_IDENTITY -> GROUP_IDENTITY_METRIC1
MIN_MATCH_IDENTITY -> GROUP_IDENTITY_METRIC2
MAX_PERCENT_MATCH -> GROUP_MATCH_METRIC1
MIN_PERCENT_MATCH -> GROUP_MATCH_METRIC2
MAX_PVALUE_MANT -> GROUP_PVALUE_MANT
MAX_PVALUE_EXP -> GROUP_PVALUE_EXP
MAX_PVALUE_MANT -> GROUP_EVALUE_MANT
MAX_PVALUE_EXP -> GROUP_EVALUE_EXP
MAX_SCORE -> GROUP_QUALITY_METRIC1
MIN_SCORE -> GROUP_QUALITY_METRIC2

Thoughts?
Chris

Chris Stoeckert, Ph.D.
Research Professor, Dept. of Genetics
1415 Blockley Hall, Center for Bioinformatics
423 Guardian Dr., University of Pennsylvania
Philadelphia, PA 19104
Ph: 215-573-4409 FAX: 215-573-3111
http://www.cbil.upenn.edu