Gene copies vs. gene paralogs - what's the difference?

Question

I'm trying to get into the theory and practice of gene copy number variation (CNV) analysis, but there is something basic confusing me, which I couldn't yet figure out. Sorry if this is a dumb/trivial question - would appreciate your help anyway.

My confusion is regarding the terms 'gene copy' and 'paralogs'. As far as I understand, paralogs are created when a gene undergoes duplication (never mind by what molecular mechanism), and then starts accumulating mutations as evolution proceeds. So, if gene X was duplicated to create another X, and then X changed to become X', are X and X' considered copies of the same gene, or are they paralogs? Is it a matter of applying some threshold on the sequence similarity between X and X', so they are considered copies up to the point where they diversify enough? Or maybe gene copies are expected to be perfect duplicates? If so, I'd guess that finding such gene pairs is very rare... Maybe it's a matter of function, so once X' gets a different function from X (neo-functionalization), it is considered a paralog? This is a rather complex and difficult to measure definition...
To make things less theoretical and more practical, lets limit the conversation to the context of whole genome sequencing analysis. Let's say that I am annotating a newly-assembled genome, and let's assume that the assembly is good enough so gene copies are not collapsed, but rather are located in their respective genomic positions. How would you look for gene copies and paralogs, and how do you make the classification?

Could anyone clarify this point for me, or refer me to relevant literature?
Thanks a lot!

Answer 1

The general term "copies" has multiple meaning with reference to genes. In diploid organisms for example, we have two copies of each gene (except those on the Y chromosome). However, these copies are called alleles. Copies that are due to duplication are called parlalogs. There is no distance criterion for differentiating copies or parlalogs. Once duplicated, there are two copies. But the consequences of paralogy would depend on differences, so the question is interesting. At the first generation after duplication, if both copies included all of the upstream regulatory elements, it would be hard to identify which of the copies was the original, since they would both be identical. But over time, if one were to evolve a new function or (more likely) accumulate frameshift substitutions, the altered one would be called the paralogous copy (relative to comparisons across species).