How many single copy gene are needed to construct the phylogenetic tree?

The one thing that strikes me is that OrthoMCL was written under David Roos supervision, which really means it was written with Toxoplasma gondii in mind, i.e. a eukaryotic parasite with notable levels of gene duplication. They tested it against Plasmodium falciparum, but that isn't surprising because both are members of the Apicomplexa. Personally I would suggest looking at a tool more orientated towards bacteria, albeit I'd like to see the output of OrthoMCL.

Prokka is a tool that might give better results, here, which is designed for bacteria, but personally I would extract the orthologes/homologues directly if the primary purpose is to contruct a phylogeny, i.e. I would parse annotated genes which are not "hypothetical".