Here are some notes on the Jackson labs database.  The schema
viewer at:
    http://www.informatics.jax.org/schema/
is helpful for understanding their database.  Galt has created
a local copy of the database using some scripts around sqsh.

The main tables we need for VisiGene are:
    GXD_Assay - Gene Expression Database Assay.  Includes info on
                in situs and other assays such as northern blots.
    MRK_Marker - Contains information on markers, many of which are genes.
    PRB_Probe  - Contains primers sometimes for RNA/DNA hybridization probes.
    GXD_InSituResultImage & GXD_InSituResult - link image to specimen
    GXD_Specimen - Describes sex, age, links to genotype
    IMG_Image - Describe image (but not how to get to image file).

Getting from gene to image:

MRK_Marker.symbol
MRK_Marker._Marker_key
GXD_Assay._Marker_key
GXD_Assay._Assay_key
[GXD_Assay._ImagePane_key  (But this is NULL some places where image exists)]
GXD_Specimen._Assay_key
GXD_Specimen._Specimen_key
GXD_InSituResult._Specimen_key
GXD_InSituResult._Result_key
GXD_InSituResultImage._Result_key
GXD_InSituResultImage._ImagePane_key
IMG_ImagePane._ImagePane_key
IMG_ImagePane._Image_key
IMG_Image._Image_key
ACC_Accession._Object_key where prefixPart="PIX:"

select MRK_Marker.symbol as gene,BIB_Refs._primary as author,
       BIB_Refs.journal as journal,GXD_Specimen.sex as sex,
       GXD_Specimen.ageMin as ageMin,GXD_Specimen.ageMax as ageMax,
       GXD_StructureName.structure as structure,
       GXD_Expression.expressed as exp, GXD_Pattern.pattern as pattern,
       IMG_ImagePane.paneLabel as pane,IMG_Image._Image_key as imageKey,
       ACC_Accession.accID as fileKey
from MRK_Marker,BIB_Refs,GXD_Assay,GXD_Expression,
     GXD_Structure,GXD_StructureName,
     GXD_Specimen,GXD_InSituResult,GXD_Pattern,
     GXD_InSituResultImage,IMG_ImagePane, IMG_Image,ACC_Accession
where MRK_Marker._Marker_key = GXD_Assay._Marker_key
  and GXD_Assay._Refs_key = BIB_Refs._Refs_key
  and GXD_Assay._Assay_key = GXD_Specimen._Assay_key
  and GXD_Assay._Assay_key = GXD_Expression._Assay_key
  and GXD_Expression._Structure_key = GXD_Structure._Structure_key
  and GXD_Structure._StructureName_key = GXD_StructureName._StructureName_key
  and GXD_Specimen._Specimen_key = GXD_InSituResult._Specimen_key
  and GXD_InSituResult._Result_key = GXD_InSituResultImage._Result_key
  and GXD_InSituResult._Pattern_key = GXD_Pattern._Pattern_key
  and GXD_InSituResultImage._ImagePane_key = IMG_ImagePane._ImagePane_key
  and IMG_ImagePane._Image_key = IMG_Image._Image_key
  and IMG_Image._Image_key = ACC_Accession._Object_key
  and ACC_Accession.prefixPart = "PIX:"
limit 1000;

select MRK_Marker.symbol,_Specimen_key,GXD_Assay._Assay_key,sex,age,ageMin,ageMax from MRK_Marker, GXD_Specimen, GXD_Assay where _Refs_key = 39558 and MRK_Marker._Marker_key = GXD_Assay._Marker_key and GXD_Specimen._Assay_key = GXD_Assay._Assay_key order by _Specimen_key limit 100;

select MRK_Marker.symbol as gene,GXD_Assay._Assay_key as assay, GXD_InSituResult._Result_key as isr,
       BIB_Refs._primary as author, BIB_Refs.journal as journal,GXD_Specimen.sex as sex,
       GXD_Specimen.ageMin as ageMin,GXD_Specimen.ageMax as ageMax,
       IMG_ImagePane.paneLabel as pane,IMG_Image._Image_key as imageKey,
       ACC_Accession.accID as fileKey
from MRK_Marker,BIB_Refs,GXD_Assay,
     GXD_Specimen,GXD_InSituResult,
     GXD_InSituResultImage,IMG_ImagePane, IMG_Image,ACC_Accession
where MRK_Marker._Marker_key = GXD_Assay._Marker_key
  and GXD_Assay._Refs_key = BIB_Refs._Refs_key
  and GXD_Assay._Assay_key = GXD_Specimen._Assay_key
  and GXD_Specimen._Specimen_key = GXD_InSituResult._Specimen_key
  and GXD_InSituResult._Result_key = GXD_InSituResultImage._Result_key
  and GXD_InSituResultImage._ImagePane_key = IMG_ImagePane._ImagePane_key
  and IMG_ImagePane._Image_key = IMG_Image._Image_key
  and IMG_Image._Image_key = ACC_Accession._Object_key
  and ACC_Accession.prefixPart = "PIX:"
  and BIB_Refs._Refs_key = 39558;
limit 1000;
    

Main table for expression is GXD_Assay.  We want cases
where the field _AssayType_key is 
   1 - RNA In Situ
   6 - Immunohistochemistry
   9 - In Situ reporter (knock in)
of which there currently are 9003 rows.

This table mostly is pointers to other tables.  Here's
the fields and what they point to:

GXD_Assay FIELDS
_Assay_key	Primary key for this table (GXD_Assay)
_AssayType_key	Key to GXD_AssayType
_Refs_key	Key to BIB_Refs which includes journal, author, etc.
_Marker_key     Key to MRK_Marker.  Table includes symbol (gene symbol) 
                when MRK_Marker._Marker_Type_key=1
_ProbePrep_key  Key to GXD_ProbePrep. References PRB_Probe, GXD_Label, 
                GXD_LabelCoverage, GXD_VisualizationMethod.
_AntibodyPrep_key Key to GXD_AntibodyPrep. References GXD_Antibody, GXD_Label.
                Either this or ProbePrep is 0.
_ImagePane_key   Key to IMG_ImagePane.  References IMG_Image, which has
_ReporterGene_key  Seems to be zero in all cases
_CreatedBy_key


Also need:

GXD_InSituResultImage FIELDS
_Result_key
_ImagePane_key

And

GXD_InSituResult FIELDS
_Result_key
_Specimen_key	Key to GXD_Specimen
_Strength_key	Key to GXD_Strength
_Pattern_key	Key to GXD_Pattern
sequenceNum
resultNote

USEFUL QUERIES
Here's a useful query to sample the GXD_Assay table:
select _AssayType_key, _Refs_key, _Marker_key, _ProbePrep_key, 
       _AntibodyPrep_key, _ImagePane_key, _ReporterGene_key, 
       _CreatedBy_key, _ModifiedBy_key 
from GXD_Assay 
where _AssayType_key in (1,6,9) 
      and (_AntibodyPrep_key != 0 || _ProbePrep_key<3000) 
order by rand() limit 10;

And the GXD_Marker table:
select _Marker_key as marker, _Organism_key as org, _Marker_Status_key as status, _Marker_Type_key as type, _CurationState_key as cura, symbol, name, chromosome as chrom, cytogeneticOffset as cyto, _CreatedBy_key from MRK_Marker order by rand() limit 10;


REPORTER GENES
74771 - lacZ
74771 - GFP