INTRO
Since release 9.1 OrthoDB provides all the data traditionally dumped in form of separate files, also as a dump of MySQL database. This should facilitate non-so-trivial queries (see examples)
IMPORT
This assumes mysql client binaries are available on the workstation and a dedicated database is created beforehand on an adequately sized MySQL server. Assuming MySQL username is odb (the password will be asked), the database name is orthodb9v1 and the server is on mysqlserver, one can use the following OS command to import the dump downloaded:
zcat ODB.sql.gz |mysql -u odb -p -h mysqlserver orthodb9v1
Importing might take an hour or so and requires less than 10GBs of disk space. After importing, one has to analyze all tables using this command at MySQL command line:
mysql> analyze table species,levels,genes,OGs,OG2genes;
USAGE
Obviously, arbitrary queries might be issued against the OrthoDB data. Here are 2 examples on how to evaluate human orthologs of a subset of mouse genes selected according to user criteria. In the first example selected are mouse kinases:
select distinct human_genes.* from
genes mouse_genes
join OG2genes using (gene_id)
join OG2genes OG2genes_2 using (cluster_id)
join genes human_genes on OG2genes_2.gene_id = human_genes.gene_id
where
match (mouse_genes.description) against ('kinase') and mouse_genes.taxon_id = 10090
and human_genes.taxon_id = 9606;

The second example shows how to evaluate human orthologs, starting from known gene identifiers (Ensembl and NCBI) via constructing on-a-fly gene list consisting of both Ensembl and NCBI subsets:
select distinct human_genes.* from
genes my_genes
join OG2genes using (gene_id)
join OG2genes OG2genes_2 using (cluster_id)
join genes human_genes on OG2genes_2.gene_id = human_genes.gene_id
where
(
my_genes.ensembl_id in (
'ENSMUSG00000028708',
'ENSMUSG00000052854',
'ENSMUSG00000105340'
) or
my_genes.gid in (
25482900,
25500718,
25485931,
25485292
))
and human_genes.taxon_id = 9606;

The later example might be easily scaled up by a scripting technique to embed user set of identifiers. Alternatively, a user-created table containing the target set of identifiers of the same origin (e.g. Uniprot) might be joined on the corresponding column (genes.uniprot_id) or SELECTed in the subquery instead of direct listing of ids.
OrthoDB genes are served with as many as possible cross references to the major databases (Esembl, Uniprot, NCBI), though some genes do lack some identifiers.