How many proteins in nature?

[rhubner at molbiol.ox.ac.uk]

hello,
 I just listened to a talk of a "drosophilist" that argued that almost every
gene/protein in higher mammals is likely to have a homologue in the fly... the
extrapolation stems from data about development regulating molecules where two
phases of duplication "amplified" the the "basic" set-up... [..."16.000 genes 
from drosophila account for some 64.000 in mammals and yield proteins with
nearly identical function"...] If such situations are plausible, then the actual
number of really "different" proteins might be cut down for the
vertebrate/invertebrate divergence, right?
 I'm not familiar with the background data... (just had same surprise than
everybody about a mouse gene involved in eye disease fostering complex eye 
formation in various regions of drosophila's body )
 On the other hand, proteins are said to be modular, but then loads of fancy
combinations could give really new, different proteins... and what about those
tiny aa differences in thermo-or psychrophilic proteins that give them another
stability/catalytic biotope whereas rest of molecule is identical?
               Just a few ideas...
greetings, Roland